from datasets import load_dataset
import streamlit as st

HF_API_TOKEN = st.secrets["HF_API_TOKEN"]
PROMPT_COLOR = "#CA437E"

ds = load_dataset("SaulLu/bloom-generations",  use_auth_token=HF_API_TOKEN)
ds = ds["train"]


def safe_text(text):
    text = text.replace("\n", "<br>")
    return f"<pre>{text}</pre>"


def prompt_markup_format(text):
    return f'<*font color="black">{text}</*font>'


def generation_markup_format(text):
    return f"<font color={PROMPT_COLOR}>{text}</pre></font>"


index_sample = st.number_input("Index of the chosen example", min_value=0, max_value=len(ds) - 1, value=0, step=1)
sample = ds[index_sample]
markdown_text = f"{prompt_markup_format(sample['prompt'])}{generation_markup_format(sample['generation'])}"
markdown_text = f"{safe_text(sample['prompt'])}{generation_markup_format(safe_text(sample['generation']))}"

st.markdown(markdown_text, unsafe_allow_html=True)