dtp-base/app/services/text.js

// text.js
// Copyright (C) 2024 DTP Technologies, LLC
// All Rights Reserved

'use strict';

import mongoose from 'mongoose';
const ChatFilter = mongoose.model('ChatFilter');

import striptags from 'striptags';
import unzalgo from 'unzalgo';
import shoetest from 'shoetest';
import diacritics from 'diacritics';
import DtpTextFilter from './lib/edit-with-vi.js';

import { SiteService, SiteError } from '../../lib/site-lib.js';

export default class TextService extends SiteService {

  static get slug () { return 'text'; }
  static get name ( ) { return 'TextService'; }

  constructor (dtp) {
    super(dtp, TextService);
  }

  async start ( ) {
    await this.loadChatFilters();
  }

  /**
   * Basic text cleaning function to remove Zalgo and tags.
   * @param {String} text The text to be cleaned
   * @returns The cleaned text
   */
  clean (text) {
    text = unzalgo.clean(text);
    text = striptags(text.trim());
    return text;
  }

  /**
   * The heavy hammer of text filtering that removes all malicious and annoying
   * things I know about as of this writing. Zalgo, tags, shoetest, diacritics,
   * and our own custom nonsense UTF-8 and Unicode filters.
   *
   * This filter is very heavy-handed and merciless.
   *
   * @param {String} text The text to be filtered
   * @returns The filtered text
   */
  filter (text) {
    if (!text || (typeof text !== 'string') || (text.length < 1)) {
      return text;
    }

    text = DtpTextFilter.filterNonsense(text);
    text = DtpTextFilter.filterGuff(text);
    text = DtpTextFilter.filterHtml(text);

    text = shoetest.simplify(text);
    text = diacritics.remove(text);

    for (const filter of this.chatFilters) {
      const regex = new RegExp(filter, 'gi');
      if (text.match(regex)) {
        this.log.alert('chat filter text match', { filter });
        throw new SiteError(403, 'Text input rejected');
      }
    }

    /*
     * Once all the stupidity has been stripped, strip the HTML
     * tags that might remain.
     */
    return this.clean(text);
  }

  async loadChatFilters ( ) {
    this.chatFilters = await ChatFilter.find().lean();
    this.chatFilters = this.chatFilters.map((filter) => filter.filter);
    this.log.debug('loading chat filters', { count: this.chatFilters.length });
  }
}