sessioncommunities.online/php/utils/sources.php

<?php
    require_once 'utils.php';

    class CommunitySources {

        private const SOURCES = array(
            'ASGL' => 'https://raw.githubusercontent.com/GNU-Linux-libre/Awesome-Session-Group-List/main/README.md',
            'LOKI' => 'https://lokilocker.com/Mods/Session-Groups/wiki/Session-Open-Groups',
            'SDIR' => 'https://session.directory/?all=groups',
            'SDIR-BASE' => 'https://session.directory/',
            'SDIR-PATTERN' => '/view_session_group_user_lokinet\.php\?id=\d+/',
            'SDIR-JSON' => 'https://session.directory/scrape.php',
            'FARK' => 'https://freearkham.cc/'
        );

        private string $contents_aggregated = "";

        private string $contents_sdir = "";

        private array $room_tags;

        /**
         * Fetches and saves known sources of Session Community join links.
         */
        public function __construct() {
            global $SOURCES;

            log_info("Requesting Awesome Session Group list...");
            $contents_asgl = file_get_contents(CommunitySources::SOURCES['ASGL']);
            log_debug($http_response_header[0]); // Supposed to be "HTTP/1.1 200 OK"

            log_info("Requesting Lokilocker Mods Open Group list...");
            $contents_loki = file_get_contents(CommunitySources::SOURCES['LOKI']);
            log_debug($http_response_header[0]); // Supposed to be "HTTP/1.1 200 OK"

            log_info("Requesting session.directory list...");
            $this->contents_sdir = file_get_contents(CommunitySources::SOURCES['SDIR-JSON']);
            log_debug($http_response_header[0]); // Supposed to be "HTTP/1.1 200 OK"

            if (!$this->sdir_process_tags()) {
                log_debug(truncate($this->contents_sdir, 50));
                log_error("Could not parse tags from session.directory.");
            }

            log_info("Requesting FreeArkham.cc list...");
            $contents_fark = file_get_contents(CommunitySources::SOURCES['FARK']);
            log_debug($http_response_header[0]); // Supposed to be "HTTP/1.1 200 OK"

            log_info('Done fetching sources.');


            $this->contents_aggregated = $contents_asgl . $contents_fark . $contents_loki . $this->contents_sdir;
        }

        private static function preprocess_tag(?string $tag) {
            $tag = trim($tag);

            if (strlen($tag) == 0) {
                return $tag;
            }

            $tag = html_sanitize(html_entity_decode($tag));

            if ($tag[0] == '#') {
                return substr($tag, 1);
            }

            return $tag;
        }

        private function sdir_process_tags(): bool {
            $entry_missing_url = false;
            $entry_missing_tags = false;

            try {
                $rooms = json_decode($this->contents_sdir, true, 512, JSON_THROW_ON_ERROR);
            } catch (JsonException) {
                return false;
            }

            foreach ($rooms as $room_entry) {
                // TODO: Check types
                if (!isset($room_entry['url'])) {
                    log_value($room_entry);
                    $entry_missing_url = true;
                    continue;
                }
                if (!isset($room_entry['tags'])) {
                    log_value($room_entry);
                    $entry_missing_tags = true;
                    continue;
                }
                $url = $room_entry['url'];
                $tags = explode(',', $room_entry['tags']);
                $room_token = url_get_token($url);
                $pubkey = url_get_pubkey($url);
                $pubkey_4 = substr($pubkey, 0, 4);
                $room_id = "$room_token+$pubkey_4";

                $tags = array_map(function(?string $tag) {
                    return CommunitySources::preprocess_tag($tag);
                }, $tags);

                $tags = array_filter(
                    $tags, function(?string $tag) {
                        return strlen($tag) != 0;
                    }
                );

                $this->room_tags[$room_id] = $tags;
            }

            if ($entry_missing_url) {
                log_error("One or more room entries from session.directory is missing the 'url' parameter.");
            }

            if ($entry_missing_tags) {
                log_error("One or more room entries from session.directory is missing the 'tags' parameter.");
            }

            return true;
        }

        /**
         * Returns all join URLs found.
         * @return string[] Join URLs.
         */
        function get_join_urls(): array {
            return array_unique(
                parse_join_links($this->contents_aggregated)
            );
        }

        /**
         * Return known tags for the given room.
         * @param string $room_id Room identifier.
         * @return string[] Array of string tags.
         */
        function get_room_tags($room_id): array {
            if (!isset($this->room_tags[$room_id])) {
                return [];
            }

            return array_slice($this->room_tags[$room_id], 0);
        }
    }
?>