import os
import json
import gzip
from string import Template
from collections import defaultdict
from fnmatch import fnmatch
from jinja2 import Environment, FileSystemLoader

env = Environment(loader=FileSystemLoader('.'))
template = env.get_template('templates/index.j2')

directory = '/data/html_data'  # Change this to your directory path
rocky_version = "8.10"

def generate_sitemap(directory):
    links = defaultdict(lambda: defaultdict(dict))
    for root, _, files in os.walk(directory):
        for file in files:
            full_filepath = os.path.join(root, file)
            filepath = full_filepath.split(rocky_version, 1)[-1]
            
            if any(fnmatch(filepath, pattern) for pattern in ['/index.html', '/links.html','/list.json*', '/sitemap*']):
                continue
            
            filepath_parts = filepath.split('/')
            package_name = filepath_parts[1]
            man_type = filepath_parts[2]
            man_type_number = man_type.lstrip('man') if man_type.startswith('man') else man_type
            command_file = filepath_parts[3]
            command = command_file.split('.html', 1)[0]
            
            if filepath.startswith('/'):
                filepath = filepath[1:]
            
            fullname = f"{package_name} - {command}({man_type_number})"
            
            links[package_name][command] = {
                "url": filepath,
                "man_type": man_type,
                "man_type_number": man_type_number,
                "fullname": fullname
            }

    return links

def generate_links_html(links):
    links_html = ""
    
    for package_name in links.keys():
        links_html += f"<h2>package_name</h2>"
        links_html += "<ul>"
        for command in links[package_name]:
            url = links[package_name][command]['url']
            man_type_number = links[package_name][command]['man_type_number']
            links_html += f"<li><a href='{url}'>{command}</a>({man_type_number})</li>"
        links_html += "</ul>"
        
    data = {
    'title': f"Rocky Man Page - {rocky_version}",
    'header_title': f"Rocky Man Page - {rocky_version}",
    'main_content': f"{links_html}"
    }   

    return template.render(data)

def convert_sitemap_to_json(links, minify=False):
    # data
    # for package_name in links.keys():
    #     for command in links[package_name]:

    #     # Add command details to sitemap
    #     sitemap[package_name][command] = {
    #         "url": filepath,
    #         "mantype": man_type,
    #         "fullname": fullname
    #     }

    if minify:
        return json.dumps(links, separators=(',', ':'))
    return json.dumps(links, indent=4)

if __name__ == "__main__":
    sitemap = generate_sitemap(directory)

    # Output the links HTML page to a file
    with open(f"{directory}/{rocky_version}/links.html", "w") as file:
        file.write(generate_links_html(sitemap))

    # Output the list JSON to a file
    with open(f"{directory}/{rocky_version}/list.json", "w") as file:
        file.write(convert_sitemap_to_json(sitemap, minify=True))
        
    # Gzip the JSON file
    with gzip.open(f"{directory}/{rocky_version}/list.json.gz", "wb") as f_out:
        f_out.write(convert_sitemap_to_json(sitemap, minify=True).encode('utf-8'))