import gradio as gr
from huggingface_hub import hf_hub_download
import json
import gzip


usernames = {}


filepath = hf_hub_download(repo_id="bigcode/the-stack-username-to-repo", filename="username_to_repo.json.gz", repo_type="dataset", revision="v1.1")
with gzip.open(filepath, 'r') as f:
    usernames["v1.1"] = json.loads(f.read().decode('utf-8'))

filepath = hf_hub_download(repo_id="bigcode/the-stack-username-to-repo", filename="username_to_repo.json.gz", repo_type="dataset")
with gzip.open(filepath, 'r') as f:
    usernames["v1.0"] = json.loads(f.read().decode('utf-8'))

text = """\
![](https://huggingface.co/spaces/bigcode/in-the-stack/resolve/main/banner.png)
**_The Stack is an open governance interface between the AI community and the open source community._**
# Stack Search By Keyword
URL: [The Stack](https://huggingface.co/datasets/bigcode/the-stack), This search engine will match your search term and find up to 100 matches by keyword for example BeatSaber.
""" + """\
"""

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 The Stack:\n\n"
        for repo in repos:
            output_md += f"_{repo}_\n\n"
    else:
        output_md += "**No**, your code is not in The Stack."
    return output_md.strip()

def check_keyword(username, version):
    output_md = ""
    maxhitcount = 100
    maxrepos = 70000000  #6M user entries * up to 18 per user
    currenthitcount=0
    currentrepos=0
    for repolist in usernames[version]:
        #print(repolist)
        repos = usernames[version][repolist]
        repo_word = "repository" if len(repos)==1 else "repositories"
        #output_md += f"**Yes**, there is code from **{len(repos)} {repo_word}** in The Stack:\n\n"
        for repo in repos:
            currentrepos += 1
            if currentrepos > maxrepos: 
                output_md += f"**Found maximum repos**, Count: **{currentrepos}** in The Stack:\n\n"
                return output_md.strip()
            if username in repo:
                currenthitcount += 1
                output_md += f"_<a href=https://github.com/{repo} target=_blank>{repo}</a>_\n\n"
                if currenthitcount > maxhitcount: 
                    output_md += f"**Found maximum hits**, Count: **{currenthitcount}** in The Stack:\n\n"
                    return output_md.strip()
    else:
        output_md += "**Searched All Repos**, Above found in The Stack."
    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(["v1.1", "v1.0"], label="The Stack version:", value="v1.1")
            username = gr.Text("", label="Keyword to match against repos e.g. BeatSaber")
            check_button = gr.Button("Check!")
            
            repos = gr.Markdown()
            
            #check_button.click(check_username, [username, version], repos)
            check_button.click(check_keyword, [username, version], repos)


demo.launch()