app.py

import gradio as gr

import json
import gzip
import urllib

usernames = {}


filepath = "username_to_repo_v1.0.json.gz"

with gzip.open(filepath, 'r') as f:
    usernames["v1.0 (from 2023-09-07 to 2023-10-06)"] = json.loads(f.read().decode('utf-8'))

filepath = "username_to_repo_v1.1.json.gz"

with gzip.open(filepath, 'r') as f:
    usernames["v1.1 (from 2023-10-07 to 2023-12-31)"] = json.loads(f.read().decode('utf-8'))

text = """\
![](https://github.com/Leolty/repobench/blob/main/assets/repobench_dark.png?raw=true)
**_RepoBench is a benchmark specifically designed for evaluating repository-level code auto-completion systems._**

# Am I in The RepoBench?

We released and maintain [RepoBench](https://arxiv.org/abs/2306.03091), built by parsing the most recent github repositories.
""" + """\

This tool lets you check if a repository under a given username is part of the RepoBench dataset. 
"""

opt_out_text_template = """\
### Opt-out

If you want your data to be removed from RepoBench for model evaluation \
open an issue with <a href="https://github.com/Leolty/repobench/issues/new?title={title}&body={body}" target="_blank">this link</a>\
"""

opt_out_issue_title = """Opt-out request for {username}"""
opt_out_issue_body = """\
I request that the following data is removed from RepoBench:

{repo_list}

_Note_: 

- If you don't want all resources to be included just remove the elements from the list above. If you would like to exclude all repositories and resources just add a single element "all" to the list. 
- If you don't want all your resources to be included in the future, just add a single element "ALL" to the list.
"""

def issue_url(username, repos):
    title = urllib.parse.quote(opt_out_issue_title.format(username=username))  
    body = urllib.parse.quote(opt_out_issue_body.format(repo_list=" - "+ "\n - ".join(repos)))
    
    opt_out_text = opt_out_text_template.format(title=title, body=body)
    
    return opt_out_text

def check_username(username, version):
    output_md = ""
    if username in usernames[version] and len(usernames[version][username])>0:
        repos = usernames[version][username]
        repo_word = "repository" if len(repos)==1 else "repositories"
        output_md += f"**Yes**, there is code from **{len(repos)} {repo_word}** in RepoBench:\n\n"
        for repo in repos:
            output_md += f"_{repo}_\n\n"

        return output_md.strip(), issue_url(username, repos)
    else:
        output_md += "**No**, your code is not in RepoBench."
        return output_md.strip(), ""

with gr.Blocks() as demo:
    with gr.Row():
        _, colum_2, _ = gr.Column(scale=1), gr.Column(scale=6), gr.Column(scale=1)
        with colum_2:
            gr.Markdown(text)
            version = gr.Dropdown([
                "v1.0 (from 2023-09-07 to 2023-10-06)",
                "v1.1 (from 2023-10-07 to 2023-12-31)"
            ], label="RepoBench version:", value="v1.1 (from 2023-10-07 to 2023-12-31)")
            username = gr.Text("", label="Your GitHub username:")
            check_button = gr.Button("Check!")
            
            repos = gr.Markdown()
            opt_out = gr.Markdown()

            
            check_button.click(check_username, [username, version], [repos, opt_out])


demo.launch()