app.py

import gradio as gr
from huggingface_hub import HfApi, hf_hub_download, Repository
from huggingface_hub.repocard import metadata_load
from gradio_client import Client
from PIL import Image, ImageDraw, ImageFont

from datetime import datetime, timezone, timedelta
import time

import os
import sys
from collections import defaultdict
import pandas as pd
import json
import shutil

api = HfApi()
HF_TOKEN = os.environ.get("HF_TOKEN")

# Public dataset repo containing the pdfs of already certified users
DATASET_REPO_URL = f"https://wseo:{HF_TOKEN}@huggingface.co/datasets/pseudolab/huggingface-krew-hackathon2023"
CERTIFIED_USERS_FILENAME = "certified.csv"

ORGANIZATION = "pseudolab"

START_DATE = datetime(2023, 10, 20, tzinfo=timezone(timedelta(hours=9)))
END_DATE = datetime(2023, 11, 10, tzinfo=timezone(timedelta(hours=9)))


def has_contributions(repo_type, hf_username, organization, likes=5):
    """
    Check if a user has contributions in the specified repository type.
    :param repo_type: A repo type supported by the Hub
    :param hf_username: HF Hub username
    :param organization: HF Hub organization
    :param likes: Minimum number of likes for a contribution to be considered
    """
    repo_list = {
        "model": api.list_models,
        "dataset": api.list_datasets,
        "space": api.list_spaces,
    }

    for repo in repo_list[repo_type](author=organization):
        if repo.likes < likes:
            continue
        commits = api.list_repo_commits(repo.id, repo_type=repo_type)
        if any(hf_username in commit.authors for commit in commits):
            return True
    return False


def get_hub_footprint(hf_username, organization):
    """
    Check the types of contributions a user has made.
    :param hf_username: HF Hub username
    :param organization: HF Hub organization
    """
    has_models = has_contributions("model", hf_username, organization)
    has_datasets = has_contributions("dataset", hf_username, organization)
    has_spaces = has_contributions("space", hf_username, organization)

    return (has_models, has_datasets, has_spaces)


def check_if_passed(hf_username):
    """
    Check if given user contributed to hackathon
    :param hf_username: HF Hub username
    """

    passed = False
    certificate_type = ""

    # If the user contributed to models, datasets and spaces then assign excellence
    has_models, has_datasets, has_spaces = get_hub_footprint(hf_username, ORGANIZATION)
    if all((has_models, has_datasets, has_spaces)):
        passed = True
        certificate_type = "excellence"
    elif any((has_models, has_datasets, has_spaces)):
        passed = True
        certificate_type = "completion"

    return passed, certificate_type


def generate_certificate(certificate_template, first_name, last_name, hf_username):
    """
    Generates certificate from the template
    :param certificate_template: type of the certificate to generate
    :param first_name: first name entered by user
    :param last_name: last name entered by user
    :param hf_username: Hugging Face Hub username entered by user
    """

    im = Image.open(certificate_template)
    d = ImageDraw.Draw(im)

    name_font = ImageFont.truetype("HeiseiMinchoStdW7.otf", 60)
    username_font = ImageFont.truetype("HeiseiMinchoStdW7.otf", 18)

    name = str(first_name) + " " + str(last_name)
    print("NAME", name)

    # Debug line name
    # d.line(((0, 419), (1000, 419)), "gray")
    # d.line(((538, 0), (538, 1400)), "gray")

    # Name
    d.text((538, 419), name, fill=(87, 87, 87), anchor="mm", font=name_font)

    # Debug line id
    # d.line(((815, 0), (815, 1400)), "gray")

    # Date of certification
    d.text((815, 327), f"HKH23-{hf_username}", fill=(117, 117, 117), font=username_font)

    pdf = im.convert("RGB")
    pdf.save("certificate.pdf")

    return im, "./certificate.pdf"


def create_initial_csv(path):
    """Create an initial CSV file with headers if it doesn't exist."""
    # Define the headers for our CSV file
    headers = [
        "hf_username",
        "first_name",
        "last_name",
        "certificate_type",
        "datetime",
        "pdf_path",
    ]
    # Create a new DataFrame with no data and these headers
    df = pd.DataFrame(columns=headers)
    # Save the DataFrame to a CSV file
    df.to_csv(path, index=False)


def add_certified_user(hf_username, first_name, last_name, certificate_type):
    """
    Add the certified user to the dataset and include their certificate PDF.
    """
    print("ADD CERTIFIED USER")
    repo = Repository(
        local_dir="data",
        clone_from=DATASET_REPO_URL,
        git_user="wseo",
        git_email="[email protected]",
    )
    repo.git_pull()

    csv_full_path = os.path.join("data", CERTIFIED_USERS_FILENAME)

    if not os.path.isfile(csv_full_path):
        create_initial_csv(csv_full_path)

    history = pd.read_csv(csv_full_path)

    # Check if this hf_username is already in our dataset:
    check = history.loc[history["hf_username"] == hf_username]
    if not check.empty:
        history = history.drop(labels=check.index[0], axis=0)

    pdfs_repo_path = os.path.join("data", "certificates")

    # Copy the PDF from its current location to the target directory in the repository
    pdf_repo_filename = f"{hf_username}.pdf"  # Create a specific name for the PDF file
    pdf_repo_path_full = os.path.join(pdfs_repo_path, pdf_repo_filename)

    # Create the pdfs directory if it doesn't exist
    os.makedirs(pdfs_repo_path, exist_ok=True)

    shutil.copy("./certificate.pdf", pdf_repo_path_full)  # Copy the file

    new_row = pd.DataFrame(
        {
            "hf_username": hf_username,
            "first_name": first_name,
            "last_name": last_name,
            "certificate_type": certificate_type,
            "datetime": time.time(),  # current time
            "pdf_path": pdf_repo_path_full[5:],  # relative path to the PDF within the repo
        },
        index=[0],
    )

    history = pd.concat([new_row, history[:]]).reset_index(drop=True)

    # Save the updated CSV
    history.to_csv(os.path.join("data", CERTIFIED_USERS_FILENAME), index=False)

    # Add the PDF and CSV changes to the repo and push
    repo.git_add()
    repo.push_to_hub(commit_message="Update certified users list and add PDF")


def create_certificate(passed, certificate_type, hf_username, first_name, last_name):
    """
    Generates certificate, adds message, saves username of the certified user
    :param passed: boolean whether the user passed enough assignments
    :param certificate_type: type of the certificate - completion or excellence
    :param hf_username: Hugging Face Hub username entered by user
    :param first_name: first name entered by user
    :param last_name: last name entered by user
    """

    if passed and certificate_type == "excellence":
        # Generate a certificate of
        certificate, pdf = generate_certificate(
            "./certificate-excellence.png", first_name, last_name, hf_username
        )
        # Add this user to our database
        add_certified_user(hf_username, first_name, last_name, certificate_type)
        # Add a message
        message = f"""
        Congratulations, you successfully completed the 2023 Hackathon 🎉! 
        Since you contributed to models, datasets, and spaces- you get a Certificate of Excellence 🎓.
        You can download your certificate below ⬇️
        https://huggingface.co/datasets/pseudolab/huggingface-krew-hackathon2023/resolve/main/certificates/{hf_username}.pdf\n
        Don't hesitate to share your certificate link above on Twitter and Linkedin (you can tag me @wonhseo, @pseudo-lab and @huggingface) 🤗
        
        """
    elif passed and certificate_type == "completion":
        # Generate a certificate of completion
        certificate, pdf = generate_certificate(
            "./certificate-completion.png", first_name, last_name, hf_username
        )
        # Add this user to our database
        add_certified_user(hf_username, first_name, last_name, certificate_type)
        # Add a message
        message = f"""
        Congratulations, you successfully completed the 2023 Hackathon 🎉!
        Since you contributed to at least one model, dataset, or space- you get a Certificate of Completion 🎓.
        You can download your certificate below ⬇️
        https://huggingface.co/datasets/pseudolab/huggingface-krew-hackathon2023/resolve/main/certificates/{hf_username}.pdf\n
        Don't hesitate to share your certificate link above on Twitter and Linkedin (you can tag me @wonhseo and @huggingface) 🤗
        You can try to get a Certificate of Excellence if you contribute to all types of repos, please don't hesitate to do so.
        """
    else:
        # Not passed yet
        certificate = Image.new("RGB", (100, 100), (255, 255, 255))
        pdf = "./fail.pdf"
        # Add a message
        message = """
          You didn't pass the minimum of one contribution to a repo *with 5 or more likes* to get a certificate of completion. 
          For more information about the certification process, refer to the hackathon page.
          If the results here differ from your contributions, make sure you moved your space to the pseudolab organization.
          """
    return certificate, message, pdf


def certification(hf_username, first_name, last_name):
    passed, certificate_type = check_if_passed(hf_username)
    certificate, message, pdf = create_certificate(
        passed, certificate_type, hf_username, first_name, last_name
    )
    print("MESSAGE", message)

    if passed:
        visible = True
    else:
        visible = False

    return message, pdf, certificate, output_row.update(visible=visible)


def make_clickable_repo(name, repo_type):
    if repo_type == "space":
        link = "https://huggingface.co/" + "spaces/" + name
    elif repo_type == "model":
        link = "https://huggingface.co/"  + name
    elif repo_type == "dataset":
        link = "https://huggingface.co/" + "datasets/" + name
    return f'<a target="_blank" href="{link}">{name.split("/")[-1]}</a>'


def make_clickable_user(user_id):
    link = "https://huggingface.co/" + user_id
    return f'<a  target="_blank" href="{link}">{user_id}</a>'


def leaderboard():
    """
    Get the leaderboard of the hackathon.
    
    The leaderboard is a Pandas DataFrame with the following columns:
    - Rank: the rank of the user in the leaderboard
    - User: the Hugging Face username of the user
    - Contributions: the list of contributions of the user (models, datasets, spaces)
    - Likes: the total number of likes of the user's contributions
    """
    repo_list = {
        'model': api.list_models,
        'dataset': api.list_datasets,
        'space': api.list_spaces
    }

    # Repos that should not be included in the leaderboard
    not_included = [
        'README',
        '2023-Hackathon-Certification',
        'huggingface-krew-hackathon2023',
        # template repos as well
    ]

    contributions = defaultdict(list)

    for repo_type in repo_list:
        for repo in repo_list[repo_type](author=ORGANIZATION):
            if repo.id.split('/')[-1] in not_included:
                continue
            commits = api.list_repo_commits(repo.id, repo_type=repo_type)
            for author in set(author for commit in commits if START_DATE < commit.created_at < END_DATE for author in commit.authors):
                contributions[author].append((repo_type, repo.id, repo.likes))

    leaderboard = []
    for user, repo_likes in contributions.items():
        repos = []
        user_likes = 0
        for repo_type, repo, likes in repo_likes:
            repos.append(make_clickable_repo(repo, repo_type))
            user_likes += likes
        leaderboard.append([make_clickable_user(user), '- ' + '\n- '.join(repos), user_likes])
    
    df = pd.DataFrame(data=leaderboard, columns=["User 👤", "Contributions 🛠️", "Likes ❤️"])
    df.sort_values(by=["Likes ❤️"], ascending=False, inplace=True)
    df.insert(0, "Rank 🏆", list(range(1, len(df) + 1)))
    
    return df


with gr.Blocks() as demo:
    gr.Markdown(
        '<img style="display: block; margin-left: auto; margin-right: auto; height: 10em;"'
            ' src="file/hfkr_logo.png"/>\n\n'
        '<h1 style="text-align: center;">Hugging Face KREW Hackathon 2023: Everyday AI</h1>'
    )
    with gr.Row():
        with gr.Column() as certificate_column:
            gr.Markdown(
                f"""
                ## Get your 2023 Hackathon Certificate 🎓
                The certification process is completely free:
                - To get a *certificate of completion*: you need to **contribute to at least one model, dataset, or space**.
                - To get a *certificate of excellence*: you need to **contribute to models, datasets, and spaces**. *(Yes, all three!)*
                
                For more information about the certification process [check the hackathon page on certification](https://pseudo-lab.github.io/huggingface-hackathon23/ko/tutorials/index.html).
                Don't hesitate to share your certificate on Twitter (tag me [@wonhseo](https://twitter.com/wonhseo) and [@huggingface](https://twitter.com/huggingface)) and on LinkedIn.
                """
            )

            hf_username = gr.Textbox(
                placeholder="wseo", label="Your Hugging Face Username (case sensitive)"
            )
            first_name = gr.Textbox(placeholder="Wonhyeong", label="Your First Name")
            last_name = gr.Textbox(placeholder="Seo", label="Your Last Name")

            check_progress_button = gr.Button(value="Check if I pass and get the certificate")
            output_text = gr.components.Textbox(label="Your Result")

            with gr.Row(visible=True) as output_row:
                output_pdf = gr.File()
                output_img = gr.components.Image(type="pil")

            check_progress_button.click(
                fn=certification,
                inputs=[hf_username, first_name, last_name],
                outputs=[output_text, output_pdf, output_img, output_row],
            )
        with gr.Column() as leaderboard_column:
            gr.Markdown(
                f"""
                ## ❤️ Leaderboard
                The leaderboard showcases your contributions for easy sharing on SNS platforms:
                
                - Event #1: *1 repo 1 share* - **share your contributions to the world!**
                - (more on the way!)
                
                For more information about the offline event [check our event-us page](https://event-us.kr/huggingfacekrew/event/72612).
                Don't hesitate to share your contributions on Twitter (tag me [@wonhseo](https://twitter.com/wonhseo) and [@huggingface](https://twitter.com/huggingface)) and on LinkedIn.

                <a class="twitter-share-button" data-size="large"
                   data-text="I'm participating in the Hugging Face KREW Hackathon 2023: Everyday AI! @wonhseo @huggingface"
                   data-url="https://huggingface.co/spaces/pseudolab/2023-Hackathon-Certification"
                   data-hashtags="huggingface,krewhackathon2023"
                   href="https://twitter.com/intent/tweet">
                Tweet</a>
                """
            )
            with gr.Row():
                repos_data = gr.components.Dataframe(
                    type="pandas", datatype=["number", "markdown", "markdown", "number"]
                )
            with gr.Row():
                data_run = gr.Button("Refresh")
                data_run.click(
                    leaderboard, outputs=repos_data
                )

    scripts = """
    async () => {
        const twitter = await import("https://platform.twitter.com/widgets.js");
        globalThis.twitter = twitter;
    }
    """
    
    demo.load(leaderboard, outputs=repos_data)
    demo.load(None, None, None, _js=scripts)

demo.launch(debug=True)