hub/gifs.js

// GIFs for Family Chat, from Tenor or GIPHY (whichever key the admin gives: TENOR_API_KEY or GIPHY_API_KEY), or with no key at all from Wikimedia Commons (MediaWiki's search API): free animated
// pictures of animals, space, science and nature rather than reaction memes. GIF_PROVIDER=tenor|giphy|wikimedia|off picks one; the default is the first of Tenor, GIPHY, Wikimedia that works. The hub does all the talking to the service, so phones never contact it:
// searches come back through here, previews are fetched here, and a GIF that is sent is downloaded and kept in the chat like any other picture.
// Everything is kept family-friendly, for everybody: only the services' G-rated results are asked for, words that could bring up adult material are refused in a search and in what comes
// back (title and tags), and only GIFs this server itself listed can be sent. A filter like this is a strong guard, not a promise: a sender or an admin can delete anything that slips through.
const TENOR = () => (process.env.TENOR_URL || 'https://tenor.googleapis.com').replace(/\/+$/, ''), GIPHY = () => (process.env.GIPHY_URL || 'https://api.giphy.com').replace(/\/+$/, '');
const WIKI = () => (process.env.WIKIMEDIA_URL || 'https://commons.wikimedia.org').replace(/\/+$/, '');
const provider = () => { const w = String(process.env.GIF_PROVIDER || '').toLowerCase(); if (w === 'off') return ''; if (w === 'wikimedia' || w === 'tenor' && process.env.TENOR_API_KEY || w === 'giphy' && process.env.GIPHY_API_KEY) return w; return process.env.TENOR_API_KEY ? 'tenor' : process.env.GIPHY_API_KEY ? 'giphy' : 'wikimedia'; };
const enabled = () => !!provider();

// whole words (and stems) that are never searched for and never shown
const BLOCK = ['sex', 'sexy', 'sexual', 'sexually', 'sexier', 'sexiest', 'porn', 'porno', 'pornography', 'pornographic', 'xxx', 'nsfw', 'nude', 'nudes', 'nudity', 'naked', 'topless', 'bottomless', 'boob', 'boobs', 'boobies', 'breast', 'breasts', 'tit', 'tits', 'titty', 'titties', 'nipple', 'nipples', 'cleavage', 'butt', 'butts', 'booty', 'twerk', 'twerking', 'thong', 'thongs', 'lingerie', 'bikini', 'bikinis', 'underwear', 'panties', 'bra', 'bras', 'strip', 'stripper', 'strippers', 'striptease', 'erotic', 'erotica', 'fetish', 'kink', 'kinky', 'bdsm', 'hentai', 'orgasm', 'orgasms', 'penis', 'vagina', 'dick', 'dicks', 'cock', 'cocks', 'pussy', 'pussies', 'cum', 'cumming', 'horny', 'slut', 'sluts', 'whore', 'hooker', 'hookers', 'masturbate', 'masturbation', 'blowjob', 'handjob', 'anal', 'milf', 'onlyfans', 'playboy', 'adult', 'adults', 'lust', 'seduce', 'seductive', 'foreplay', 'grope', 'groping', 'fuck', 'fucking', 'fucker', 'fucked', 'shit', 'bitch', 'bitches', 'asshole', 'ass', 'drugs', 'cocaine', 'marijuana', 'gore', 'suicide'];
const BLOCKED = new Set(BLOCK);
// Commons files carry categories, so those are checked too, against a few more phrases (a category is a phrase, not a word)
const RISKY = /nudity|nude|naked|sexual|sexuality|erotic|pornograph|genital|fetish|bdsm|topless|underwear|swimwear|bikini|lingerie|breast|buttock|nipple|prostitut|obscene|sex education|human sexuality|anatomy of the|violence|corpse|dead bod|execution|torture|massacre|gore|suicide|drug use|cannabis|cocaine|heroin/i;
/** Does this text (a search, a title, tags) contain a word that is not allowed? Whole words only, so "class" and "assistant" are fine. */
const unsafe = text => (String(text || '').toLowerCase().match(/[a-z0-9]+/g) || []).some(w => BLOCKED.has(w) || BLOCKED.has(w.replace(/s$/, '')));

// only GIFs this server has just listed can be sent or previewed (so the download address cannot be chosen by the sender)
const known = new Map(); // url -> expiry
const remember = urls => { const t = Date.now() + 3600e3; for (const u of urls) known.set(u, t); if (known.size > 4000) for (const [k, v] of known) if (v < Date.now()) known.delete(k); };
const isKnown = u => (known.get(u) || 0) > Date.now();
const MEDIA = /^https:\/\/(media[0-9]*\.tenor\.com|media[0-9]*\.giphy\.com|i\.giphy\.com|c\.tenor\.com|upload\.wikimedia\.org)\//i;
const mediaOk = u => { try { const x = new URL(u); return MEDIA.test(u) || String(process.env.GIF_MEDIA_HOSTS || '').split(',').filter(Boolean).includes(x.host); } catch { return false; } };

async function getJson(url) {
  const r = await fetch(url, { headers: { accept: 'application/json', 'user-agent': 'NE-Hub/1.0 (family chat)' }, signal: AbortSignal.timeout(8000) });
  if (r.status === 403 || r.status === 401) throw Object.assign(new Error('The GIF service refused the key. Check it in the stack settings.'), { code: 502 });
  if (r.status === 429) throw Object.assign(new Error('The GIF service is busy. Try again in a minute.'), { code: 429 });
  if (!r.ok) throw Object.assign(new Error('The GIF service did not answer (' + r.status + ')'), { code: 502 });
  return r.json();
}
/** Search (or, with no words, what is popular). Returns { items: [{ title, w, h, preview, url }], next }. */
async function search(q, pos) {
  q = String(q || '').trim().slice(0, 60);
  if (q && unsafe(q)) throw Object.assign(new Error('That search is not available. Try something else.'), { code: 400 });
  const n = 24; let items = [], next = '';
  if (provider() === 'tenor') {
    const key = encodeURIComponent(process.env.TENOR_API_KEY), base = `${TENOR()}/v2/${q ? 'search?q=' + encodeURIComponent(q) + '&' : 'featured?'}key=${key}&client_key=nehub&limit=${n}&contentfilter=high&media_filter=tinygif,gif${pos ? '&pos=' + encodeURIComponent(pos) : ''}`;
    const j = await getJson(base); next = j.next || '';
    items = (j.results || []).map(x => { const t = (x.media_formats || {}).tinygif, g = (x.media_formats || {}).gif || t; return t && g ? { title: x.content_description || '', tags: (x.tags || []).join(' '), w: (t.dims || [])[0] || 0, h: (t.dims || [])[1] || 0, preview: t.url, url: g.url } : null; });
  } else if (provider() === 'giphy') {
    const off = Math.max(0, parseInt(pos, 10) || 0), key = encodeURIComponent(process.env.GIPHY_API_KEY), base = `${GIPHY()}/v1/gifs/${q ? 'search?q=' + encodeURIComponent(q) + '&' : 'trending?'}api_key=${key}&limit=${n}&offset=${off}&rating=g&bundle=messaging_non_clips`;
    const j = await getJson(base); next = (j.data || []).length >= n ? String(off + n) : '';
    items = (j.data || []).filter(x => !x.rating || x.rating === 'g').map(x => { const t = (x.images || {}).fixed_width_small || (x.images || {}).fixed_width, g = (x.images || {}).fixed_width || (x.images || {}).original; return t && g && t.url && g.url ? { title: x.title || '', tags: '', w: +t.width || 0, h: +t.height || 0, preview: t.url, url: g.url } : null; });
  } else if (provider() === 'wikimedia') { // Wikimedia Commons: animated GIFs only, chosen by the search itself; no rating exists, so the word and category checks do the work
    const off = Math.max(0, parseInt(pos, 10) || 0), term = (q || 'animal') + ' filemime:image/gif';
    const url = `${WIKI()}/w/api.php?action=query&format=json&generator=search&gsrnamespace=6&gsrlimit=${n}&gsroffset=${off}&gsrsearch=${encodeURIComponent(term)}&prop=imageinfo&iiprop=url|size|mime|extmetadata&iiurlwidth=320&iiextmetadatafilter=ImageDescription|Categories|ObjectName`;
    const j = await getJson(url); next = j.continue && j.continue.gsroffset != null ? String(j.continue.gsroffset) : '';
    items = Object.values((j.query || {}).pages || {}).sort((a, b) => (a.index || 0) - (b.index || 0)).map(pg => {
      const ii = (pg.imageinfo || [])[0]; if (!ii || ii.mime !== 'image/gif' || !ii.thumburl) return null; const em = ii.extmetadata || {}, plain = v => String(v && v.value || '').replace(/<[^>]*>/g, ' ').replace(/\s+/g, ' ').trim();
      const title = String(pg.title || '').replace(/^File:/, '').replace(/\.gif$/i, '').replace(/[_]+/g, ' '), cats = plain(em.Categories).replace(/\|/g, ' · '), desc = plain(em.ImageDescription).slice(0, 300);
      if (RISKY.test(cats) || RISKY.test(title) || RISKY.test(desc)) return null; return { title: plain(em.ObjectName) || title, tags: cats + ' ' + desc, w: ii.thumbwidth || 0, h: ii.thumbheight || 0, preview: ii.thumburl, url: ii.thumburl };
    });
  } else throw Object.assign(new Error('GIFs are not set up on this server'), { code: 503 });
  items = items.filter(x => x && mediaOk(x.preview) && mediaOk(x.url) && !unsafe(x.title) && !unsafe(x.tags)).map(({ tags, ...x }) => x);
  remember(items.flatMap(x => [x.preview, x.url])); return { items, next };
}
/** The bytes of a GIF this server listed (a preview or the full one), size-limited. */
async function fetchBytes(url, max = 8 * 1024 * 1024) {
  if (!isKnown(url) || !mediaOk(url)) throw Object.assign(new Error('That GIF is not available'), { code: 400 });
  const r = await fetch(url, { headers: { 'user-agent': 'NE-Hub/1.0 (family chat)' }, signal: AbortSignal.timeout(15000), redirect: 'error' });
  if (!r.ok) throw Object.assign(new Error('The GIF could not be downloaded'), { code: 502 });
  const type = String(r.headers.get('content-type') || '').split(';')[0].trim().toLowerCase(); if (!/^image\/(gif|webp|png|jpeg)$/.test(type)) throw Object.assign(new Error('That is not a picture'), { code: 502 });
  const buf = Buffer.from(await r.arrayBuffer()); if (buf.length > max) throw Object.assign(new Error('That GIF is too large'), { code: 413 }); return { buf, type };
}
module.exports = { enabled, provider, search, fetchBytes, unsafe, isKnown, remember };