galactic-bot/structure/client/components/observers/Automoderation.js

const { inspect } = require('util');
const similarity = require('similarity');
const { stripIndents } = require('common-tags');

const { Observer, BinaryTree } = require('../../../interfaces');
const { FilterUtil, FilterPresets, Util } = require('../../../../util');
const { Warn, Mute, Kick, Softban, Ban } = require('../../../moderation/infractions');

const CONSTANTS = {
    Infractions: {
        WARN: Warn,
        MUTE: Mute,
        KICK: Kick,
        SOFTBAN: Softban,
        BAN: Ban
    }
};

// TODO: 
// Clean up commented out code once testing of new code is done
// Implement missing automod features


module.exports = class AutoModeration extends Observer {

    constructor(client) {

        super(client, {
            name: 'autoModeration',
            priority: 1,
            disabled: false
        });

        this.hooks = [
            ['message', this.filterWords.bind(this)],
            ['messageUpdate', this.filterWords.bind(this)],
            ['message', this.flagMessages.bind(this)],
            ['messageUpdate', this.flagMessages.bind(this)],
            ['message', this.filterLinks.bind(this)],
            ['messageUpdate', this.filterLinks.bind(this)],
            ['message', this.filterInvites.bind(this)],
            ['messageUpdate', this.filterInvites.bind(this)],
            ['message', this.filterMentions.bind(this)]
        ];

        this.whitelist = new BinaryTree(this.client, FilterPresets.whitelist);

    }

    _moderate(action, guild, channel, member, moderationPoints, silent, reason, filterResult, message) {
        this.client.moderationManager.autoModerate({ guild, action, message, reason, silent, data: { filterResult } });

        /* this.client.moderationManager.handleInfraction(
            CONSTANTS.Infractions[action.type],
            { // Hacky patched together message object with just the required stuff for modmanager to work
                guild,
                member: guild.me,
                channel,
                arguments: {
                    force: {
                        value: action.force
                    },
                    points: {
                        value: action.points || moderationPoints.points[action.type]
                    },
                    expiration: {
                        value: action.expiration || moderationPoints.expirations[action.type]
                    },
                    silent
                },
                format: guild.format.bind(guild),
                // eslint-disable-next-line no-empty-function
                respond: () => { }
            },
            {
                targets: [member],
                reason,
                duration: action.duration,
                data: {
                    filterResult
                }
            }
        );*/
    }

    async filterWords(message, edited) {

        const { guild, author, channel, command } = message;
        if (!guild || author.bot) return;

        const member = message.member || await guild.members.fetch(author.id).catch();
        const settings = await guild.settings();
        const { wordFilter: setting, moderationPoints } = settings;
        const { bypass, ignore, enabled, silent, explicit, fuzzy, regex, whitelist, actions } = setting;
        const roles = member?.roles.cache.map((r) => r.id) || [];

        if (!enabled || roles.some((r) => bypass.includes(r)) || ignore.includes(channel.id)) return;

        if (command?.name === 'settings') {
            // NOTE: probably needs a more permanent solution
            const result = await this.client.registry.components.get('inhibitor:permissions').execute(message, message.command);
            if (!result.error) return;
        }

        // Which message obj to work with
        const msg = edited || message;
        if (!msg.content) return;
        let log = `Message filter debug:`;
        log += `\nPre norm: ${msg.cleanContent}`;
        let content = null;
        try { // NOTE: Remove try-catch after debug
            content = FilterUtil.normalise(Util.removeMarkdown(msg.cleanContent));
        } catch (err) {
            this.client.logger.debug(log);
            return;
        }
        log += `\nNormalised: ${content}`;

        const catcher = (ln) => {
            return () => this.client.logger.debug(`Issue with promise on line ${ln}`);
        };

        // match: what was matched | 
        // matched: did it match at all ? | 
        // matcher: what gets shown in the message logs | 
        // _matcher: locally used variable for which word in the list triggered it | 
        // type: which detection type matched it
        let filterResult = { match: null, matched: false, matcher: null, _matcher: null, preset: false };
        const words = content.toLowerCase().split(' ').filter((elem) => elem.length);
        // Remove any potential bypass characters
        //const _words = words.map((word) => word.replace(/[.'*_?+"#%&=-]/gu, ''));

        // 1. Filter for preset lists
        // CHANGED BEHAVIOUR OF PRESETS, THEY NOW USE THE REGEX PART OF THE FILTER
        // if (presets.length) {
        //     for (const preset of presets) {

        //         const text = _words.join('').replace(/\s/u, ''); //Also check for spaced out words, ex "f u c k"
        //         //Combine array of presets to one expression
        //         const _regex = new RegExp(`(${FilterPresets[preset].join(')|(')})`, 'ui');
        //         const match = content.match(_regex) || text.length === words.length ? text.match(_regex) : null;
        //         if (!match) continue;
        //         this.client.logger.debug(`\nMessage matched with "${preset}" preset list.\nMatch: ${match[0]}\nFull content: ${content}`);
        //         filterResult = {
        //             match: match[0],
        //             matched: true,
        //             matcher: preset,
        //             preset,
        //             type: 'preset'
        //         };
        //         break;

        //     }
        // }

        // 2. Filter explicit - no bypass checking (unless you count normalising the text, i.e. emoji letters => normal letters)
        if (explicit.length && !filterResult.matched) {

            //filterResult = FilterUtil.filterExplicit(words, explicit);
            // if(filterResult)

            for (const word of explicit) {
                //Do it like this instead of regex so it doesn't match stuff like Scunthorpe with cunt
                if (words.some((_word) => _word === word)) {
                    log += `\nMessage matched with "${word}" in the explicit list.\nFull content: ${content}`;
                    filterResult = {
                        match: word,
                        matched: true,
                        matcher: 'explicit',
                        _matcher: word,
                        type: 'explicit'
                    };
                    break;
                }

            }

        }

        // 3. Filter regex
        if (regex.length && !filterResult.matched) {

            for (const reg of regex) {

                const match = content.match(new RegExp(`(?:^|\\s)(${reg})`, 'iu')); // (?:^|\\s) |un

                if (match) {
                    //log += `\next reg: ${tmp}`;
                    const fullWord = words.find((word) => word.includes(match[1]));

                    let inWL = false;
                    try { // This is for debugging only
                        inWL = this.whitelist.find(fullWord);
                    } catch (err) {
                        this.client.logger.debug(fullWord, match[0], words);
                    }
                    if (inWL || whitelist.some((word) => word === fullWord)) continue;

                    log += `\nMessage matched with "${reg}" in the regex list.\nMatch: ${match[0]}, Full word: ${fullWord}\nFull content: ${content}`;
                    filterResult = {
                        match: fullWord,
                        matched: true,
                        _matcher: match[0].toLowerCase(),
                        matcher: `Regex: __${reg}__`,
                        type: 'regex'
                    };
                    break;
                }

            }

        }

        // 4. Filter fuzzy
        if (fuzzy.length && !filterResult.matched) {

            const text = words.join('').replace(/\s/u, '');
            const threshold = 0.93 - 0.165 * Math.log(text.length);

            outer:
            for (const _word of fuzzy) {

                for (const word of words) {
                    const sim = similarity(word, _word);
                    const threshold = 0.93 - 0.165 * Math.log(word.length);
                    if (sim >= threshold && Math.abs(_word.length - word.length) < 3) {
                        if (this.whitelist.find(word) || whitelist.some((w) => w === word) && sim < 1) continue;
                        log += `\nMessage matched with "${_word}" in fuzzy.\nMatched word: ${word}\nFull content: ${content}\nSimilarity: ${sim}\nThreshold: ${threshold}`;
                        filterResult = {
                            match: word,
                            matched: true,
                            _matcher: _word,
                            matcher: `fuzzy [\`${_word}\`, \`${sim}\`, \`${threshold}\`]`,
                            type: 'fuzzy'
                        };
                        break outer;
                    }

                }

                const sim = similarity(text, _word);
                if (sim >= threshold && Math.abs(_word.length - text.length) < 3) {
                    if (this.whitelist.find(text) || whitelist.some((w) => w === text) && sim < 1) continue;
                    log += `\nMessage matched with "${_word}" in fuzzy.\nMatched word: ${text}\nFull content: ${content}\nSimilarity: ${sim}\nThreshold: ${threshold}`;
                    filterResult = {
                        match: text,
                        matched: true,
                        _matcher: _word,
                        matcher: `fuzzy [\`${_word}\`, \`${sim}\`, \`${threshold}\`]`,
                        type: 'fuzzy'
                    };
                    break;
                }

                //this.client.logger.debug(`Message did not match with "${_word}" in fuzzy.\nFull content: ${content}\nSimilarity: ${sim}\nThreshold: ${threshold}`);

            }

        }

        // 5. Remove message, inline response and add a reason to msg object
        if (!filterResult.matched) return;
        msg.filtered = filterResult;
        log += `\nFilter result: ${inspect(filterResult)}`;
        if (!silent) {
            const res = await this.client.rateLimiter.limitSend(msg.channel, msg.format('W_FILTER_DELETE', { user: author.id }), undefined, 'wordFilter').catch(catcher(255));
            //const res = await msg.formattedRespond('W_FILTER_DELETE', { params: { user: author.id } });
            if (res) res.delete({ timeout: 10000 }).catch(catcher(257));
        }

        // 6. Automated actions
        if (actions.length) {

            let action = actions.find((act) => {
                return act.trigger.includes(filterResult._matcher);
            });
            if (!action) action = actions.find((act) => {
                return act.trigger === filterResult.type;
            });
            if (!action) action = actions.find((act) => {
                return act.trigger === 'generic';
            });

            if (!action) {
                this.client.logger.debug(log);
                this.client.rateLimiter.queueDelete(msg.channel, msg).catch(catcher(275));
            }
            
            msg.filtered.sanctioned = true;
            this.client.rateLimiter.queueDelete(msg.channel, msg).catch(catcher(279));
            this.client.logger.debug(log + '\nSanctioned');
            
            // NOTE: this will have to be changed whenever the moderation manager is finished and properly supports sth like this
            this._moderate(action, guild, channel, member, moderationPoints, silent, msg.format('W_FILTER_ACTION'), filterResult, message);

        } else {
            this.client.rateLimiter.queueDelete(msg.channel, msg).catch(catcher(286));
            this.client.logger.debug(log);
        }

    }

    async flagMessages(message, edited) {

        const { guild, author, channel } = message;
        if (!guild || author.bot) return;

        const member = message.member || await guild.members.fetch(author.id).catch();
        const settings = await guild.settings();
        const { wordWatcher: setting } = settings;
        const { words, bypass, ignore, channel: _logChannel } = setting;
        const roles = member?.roles.cache.map((r) => r.id) || [];

        if (!_logChannel || words.length === 0 || roles.some((r) => bypass.includes(r.id)) || ignore.includes(channel.id)) return;

        const logChannel = await guild.resolveChannel(_logChannel);
        const msg = edited || message;
        if (!msg.content) return;
        const content = FilterUtil.normalise(msg.cleanContent);
        let match = null;

        for (const reg of words) {
            
            match = content.match(new RegExp(`(?:^|\\s)(${reg})`, 'iu'));

            if (match) break;

        }

        if (!match) return;

        const context = channel.messages.cache.sort((m1, m2) => m2.createdTimestamp - m1.createdTimestamp).first(5);
        const embed = {
            title: `⚠️ Word trigger in **#${channel.name}**`,
            description: stripIndents`
            **[Jump to message](${msg.link})**
            `, // ** User:** <@${ author.id }>
            color: 15120384,
            fields: context.reverse().reduce((acc, val) => {
                const text = val.content.length ? Util.escapeMarkdown(val.content).replace(match[1], '**__$&__**') : '**NO CONTENT**';
                acc.push({
                    name: `${val.author.tag} (${val.author.id}) - ${val.id}`,
                    value: text.length < 1024 ? text : text.substring(0, 1013) + '...'
                });
                if (text.length > 1024) acc.push({
                    name: `\u200b`,
                    value: '...' + text.substring(1013, 2034)
                });
                return acc;
            }, [])
        };

        const sent = await logChannel.send({ embed }).catch((err) => {
            this.client.logger.error('Error in message flag:\n' + err.stack);
        });

    }

    async filterLinks(message, edited) {

        const { guild, author, channel } = message;
        if (!guild || author.bot) return;

        const member = message.member || await guild.members.fetch(author.id).catch();
        const { resolver } = this.client;
        const settings = await guild.settings();
        const { linkFilter: setting, moderationPoints } = settings;
        const { bypass, ignore, actions, silent, enabled, blacklist, whitelist, mode } = setting;
        if (!enabled) return;
        const roles = member?.roles.cache.map((r) => r.id) || [];

        if (roles.some((r) => bypass.includes(r.id)) || ignore.includes(channel.id)) return;

        const msg = edited || message;
        if (!msg.content) return;
        const content = msg.content.split('').join(''); //Copy the string...
        const linkRegG = /(https?:\/\/(www\.)?)?(?<domain>([a-z0-9-]{1,63}\.)?([a-z0-9-]{2,63})(\.[a-z0-9-]{2,63})(\.[a-z0-9-]{2,63})?)(\/\S*)?/iug;
        const linkReg = /(https?:\/\/(www\.)?)?(?<domain>([a-z0-9-]{1,63}\.)?([a-z0-9-]{2,63})(\.[a-z0-9-]{2,63})(\.[a-z0-9-]{2,63})?)(\/\S*)?/iu;
        let matches = content.match(linkRegG);
        if (!matches) matches = content.replace(/\s/u, '').match(linkRegG);
        if (!matches) return;
        let remove = false;
        const filterResult = {};
        const _whitelist = mode === 'whitelist';

        for (const match of matches) {
            const { domain } = match.match(linkReg).groups;

            // eslint-disable-next-line brace-style
            if (!_whitelist && blacklist.some((dom) => { return dom.includes(domain) || domain.includes(dom); })) {
                filterResult.match = domain;
                filterResult.matcher = 'link blacklist';
                remove = true;
                break;
            } else if (_whitelist) {
                // eslint-disable-next-line brace-style
                if (whitelist.some((dom) => { return dom.includes(domain) || domain.includes(dom); })) continue;
                const valid = await resolver.validateDomain(domain);
                if (!valid) continue;

                filterResult.match = domain;
                filterResult.matcher = 'link whitelist';
                remove = true;
                break;
            } 

        }

        if (!remove) return;
        msg.filtered = filterResult;
        if (!silent) {
            const res = await this.client.rateLimiter.limitSend(msg.channel, msg.format('L_FILTER_DELETE', { user: author.id }), undefined, 'linkFilter');
            //const res = await msg.formattedRespond(`L_FILTER_DELETE`, { params: { user: author.id } });
            if (res) res.delete({ timeout: 10000 });
        }

        if (actions.length) {

            let action = actions.find((act) => {
                return act.trigger.includes(filterResult.match);
            });
            if (!action) action = actions.find((act) => {
                return act.trigger === mode;
            });
            if (!action) action = actions.find((act) => {
                return act.trigger === 'generic';
            });

            if (!action) return msg.delete();

            msg.filtered.sanctioned = true;
            this.client.rateLimiter.queueDelete(msg.channel, msg);
            //msg.delete();

            this._moderate(action, guild, channel, member, moderationPoints, silent, msg.format('L_FILTER_ACTION'), filterResult, message);
            
        } else this.client.rateLimiter.queueDelete(msg.channel, msg); //msg.delete();

    }

    async filterInvites(message, edited) {

        const { guild, author, channel } = message;
        if (!guild || author.bot) return;

        const member = message.member || await guild.members.fetch(author.id).catch();
        const settings = await guild.settings();
        const { inviteFilter: setting, moderationPoints } = settings;
        const { bypass, ignore, actions, silent, enabled } = setting;
        if (!enabled) return;
        const roles = member?.roles.cache.map((r) => r.id) || [];

        if (roles.some((r) => bypass.includes(r.id)) || ignore.includes(channel.id)) return;

        const msg = edited || message;
        const { content } = msg;
        if (!content) return;

        const reg = /((discord)?\s?\.?\s?gg\s?|discord(app)?\.com\/invite)\/\s?(?<code>[a-z0-9]+)/iu;
        const match = content.match(reg);
        if (!match) return;

        const result = await guild.checkInvite(match.groups.code);
        if (!result) { // Doesn't resolve to the origin server

            let action = null;
            if (actions.length) [action] = actions; 
            
            msg.filtered = {
                match: match[0],
                matcher: 'invites'
            };
            if (!action) return this.client.rateLimiter.queueDelete(msg.channel, msg); //msg.delete();
            if (!silent) {
                const res = await this.client.rateLimiter.limitSend(msg.channel, msg.format('I_FILTER_DELETE', { user: author.id }), undefined, 'inviteFilter');
                if (res) res.delete({ timeout: 10000 });
            }
            msg.filtered.sactioned = true;
            this.client.rateLimiter.queueDelete(msg.channel, msg);

            this._moderate(action, guild, channel, member, moderationPoints, silent, msg.format('I_FILTER_ACTION'), { filtered: msg.filtered }, message);

        }

    }

    async filterMentions(message) {

        const { guild, author, channel } = message;
        if (!guild || author.bot) return;

        const member = message.member || await guild.members.fetch(author.id).catch();
        const settings = await guild.settings();
        const { mentionFilter: setting, moderationPoints } = settings;
        const { bypass, ignore, enabled, silent, unique, mentions, actions } = setting;
        const roles = member?.roles.cache.map((r) => r.id) || [];

        if (!enabled || roles.some((r) => bypass.includes(r.id)) || ignore.includes(channel.id)) return;

        const reg = /<@!?[0-9]{18,22}>/gu;
        const { content } = message;
        if (!content) return;
        //const mentions = content.match(reg);

    }

};