app.py

import io
import json
import os

import gradio as gr
import markdown
import pandas as pd
from gchar.games.dispatch.access import get_character
from gchar.resources.pixiv import get_pixiv_keywords, get_pixiv_posts
from gchar.resources.sites import list_available_sites, get_site_tag
from huggingface_hub import hf_hub_url

from character import get_ch_name
from huggingface import get_hf_fs

hf_fs = get_hf_fs()


def query(chr_name):
    ch = get_character(chr_name, allow_fuzzy=True)

    # get character info
    info_columns = ['Property', 'Value']
    info_data = []
    info_data.append(('Index', ch.index))
    ennames = [str(enname) for enname in ch.ennames]
    if ennames:
        info_data.append(('EN Name', ', '.join(ennames)))
    cnnames = [str(cnname) for cnname in ch.cnnames]
    if cnnames:
        info_data.append(('CN Name', ', '.join(cnnames)))
    jpnames = [str(jpname) for jpname in ch.jpnames]
    if jpnames:
        info_data.append(('JP Name', ', '.join(jpnames)))
    if hasattr(ch, 'krnames'):
        krnames = [str(krname) for krname in ch.krnames]
        if krnames:
            info_data.append(('KR Name', ', '.join(krnames)))
    info_data.append(('Sex', ch.gender.name))
    info_data.append(('Source', ch.__official_name__))
    info_df = pd.DataFrame(columns=info_columns, data=info_data)

    # get skins
    skin_dir = f'datasets/deepghs/game_character_skins/{ch.__game_name__}/{ch.index}'
    meta_json = f'{skin_dir}/.meta.json'
    skin_urls = []
    if hf_fs.exists(meta_json):
        meta = json.loads(hf_fs.read_text(meta_json))
        for item in meta['files']:
            skin_urls.append(hf_hub_url(
                'deepghs/game_character_skins',
                filename=f'{ch.__game_name__}/{ch.index}/{item["name"]}',
                repo_type='dataset',
            ))

    # get repo info
    repo = f'CyberHarem/{get_ch_name(ch)}'
    with io.StringIO() as sf:
        if hf_fs.exists(f'{repo}/meta.json'):
            model_url = f'https://huggingface.co/{repo}'
            print(f'Model: [{model_url}]({model_url})', file=sf)
        else:
            print(f'Model not found.', file=sf)
        print(file=sf)

        if hf_fs.exists(f'datasets/{repo}/dataset-raw.zip'):
            ds_url = f'https://huggingface.co/datasets/{repo}'
            print(f'Dataset: [{ds_url}]({ds_url})', file=sf)
        else:
            print('Dataset not found.', file=sf)
        print(file=sf)
        html = markdown.markdown(sf.getvalue())

    # get tags on all sites
    tags_columns = ['Site', 'Posts', 'Tag']
    tags_data = []
    tags_data.append(('Pixiv (ALL)', get_pixiv_posts(ch)[0], get_pixiv_keywords(ch)))
    tags_data.append(('Pixiv (R18)', get_pixiv_posts(ch)[1], get_pixiv_keywords(ch, includes=['R-18'])))
    for site in list_available_sites():
        tag_retval = get_site_tag(ch, site, with_posts=True, sure_only=True)
        if tag_retval is not None:
            tag_name, tag_cnt = tag_retval
            tags_data.append((site, tag_cnt, tag_name))
    tags_data = sorted(tags_data, key=lambda x: (-x[1], x[0]))
    tags_df = pd.DataFrame(columns=tags_columns, data=tags_data)

    return info_df, skin_urls, html, tags_df


if __name__ == '__main__':
    with gr.Blocks() as demo:
        gr_input = gr.Textbox(
            label='Character Name',
            placeholder='Enter name or alias of the character.'
        )
        gr_submit = gr.Button(value='Find My Waifu', variant='primary')

        with gr.Row():
            with gr.Column():
                with gr.Row():
                    gr_info = gr.DataFrame(label='Character Info')
                with gr.Row():
                    gr_skins = gr.Gallery(label='Skins')

            with gr.Column():
                with gr.Row():
                    gr_html = gr.HTML(label='Entry of Model and Dataset', value='(N/A)')
                with gr.Row():
                    gr_tags = gr.DataFrame(label='Character Tags')

        gr_submit.click(
            query,
            inputs=[
                gr_input,
            ],
            outputs=[
                gr_info,
                gr_skins,
                gr_html,
                gr_tags,
            ]
        )

    demo.queue(os.cpu_count()).launch()