halbot 1994.1.8 → 1995.1.2

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/lib/hal.mjs CHANGED
@@ -1,46 +1,25 @@
1
- import {
2
- bot, callosum, dbio, media, speech, storage, uoid, utilitas, web,
3
- } from 'utilitas';
4
-
5
- import { basename, join } from 'path';
1
+ import { alan, bot, callosum, dbio, storage, utilitas, } from 'utilitas';
2
+ import { join } from 'path';
6
3
  import { parseArgs as _parseArgs } from 'node:util';
7
4
  import { readdirSync } from 'fs';
8
5
 
9
- const _NEED = ['lorem-ipsum', 'mime'];
10
6
  // 👇 https://core.telegram.org/bots/faq#my-bot-is-hitting-limits-how-do-i-avoid-this
11
- const table = 'utilitas_hal_events';
12
- const [PRIVATE_LIMIT, GROUP_LIMIT] = [60 / 60, 60 / 20].map(x => x * 1000);
13
- const log = (c, o) => utilitas.log(c, import.meta.url, { time: 1, ...o || {} });
14
- const [end, parse_mode] = [bot.end, bot.parse_mode];
15
- const normalizeKey = chatId => `${HALBOT}_SESSION_${chatId}`;
16
- const lines2 = arr => bot.lines(arr, '\n\n');
17
- const uList = arr => bot.lines(arr.map(x => `- ${x}`));
18
- const oList = arr => bot.lines(arr.map((v, k) => `${k + 1}. ${v}`));
19
- const isMarkdownError = e => e?.description?.includes?.("can't parse entities");
20
- const getFile = async (id, op) => (await web.get(await getFileUrl(id), op)).content;
21
- const compact = (str, op) => utilitas.ensureString(str, { ...op || {}, compact: true });
22
- const compactLimit = (str, op) => compact(str, { ...op || {}, limit: 140 });
23
- const getKey = s => s?.toLowerCase?.()?.startsWith?.('http') ? 'url' : 'source';
24
- const SEARCH_LIMIT = 10;
25
-
26
7
  const [ // https://limits.tginfo.me/en
27
- HELLO, GROUP, PRIVATE, CHANNEL, MENTION, CALLBACK_LIMIT, API_ROOT,
28
- jsonOptions, sessions, HALBOT, COMMAND_REGEXP, COMMAND_LENGTH,
29
- COMMAND_LIMIT, COMMAND_DESCRIPTION_LENGTH, bot_command, EMOJI_SPEECH,
30
- EMOJI_LOOK, EMOJI_BOT, logOptions, ON, OFF,
8
+ HELLO, GROUP, PRIVATE, CHANNEL, MENTION, jsonOptions, COMMAND_LIMIT, ON,
9
+ OFF, BOT_COMMAND, CHECK, logOptions, COMMAND_LENGTH,
10
+ COMMAND_DESCRIPTION_LENGTH, RELEVANCE, SEARCH_LIMIT, SUB_LIMIT,
31
11
  ] = [
32
- 'Hello!', 'group', 'private', 'channel', 'mention', 30,
33
- 'https://api.telegram.org/', { code: true, extraCodeBlock: 1 }, {},
34
- 'HALBOT', /^\/([a-z0-9_]+)(@([a-z0-9_]*))?\ ?(.*)$/sig, 32, 100, 256,
35
- 'bot_command', '👂', '👀', '🤖', { log: true }, 'on', 'off',
12
+ 'Hello!', 'group', 'private', 'channel', 'mention',
13
+ { code: true, extraCodeBlock: 1 }, 100, 'on', 'off', 'bot_command',
14
+ '☑️', { log: true }, 32, 256, 0.2, 10, 200, // Google Rerank limit
36
15
  ];
37
16
 
38
- const [BUFFER_ENCODE, BINARY_STRINGS] = [{ encode: storage.BUFFER }, [OFF, ON]];
39
-
40
- const KNOWN_UPDATE_TYPES = [
41
- 'callback_query', 'channel_post', 'edited_message', 'message',
42
- 'my_chat_member', // 'inline_query',
43
- ];
17
+ const table = 'utilitas_hal_events';
18
+ const log = (c, o) => utilitas.log(c, import.meta.url, { time: 1, ...o || {} });
19
+ const [end] = [bot.end];
20
+ const uList = arr => bot.lines(arr.map(x => `- ${x}`));
21
+ const oList = arr => bot.lines(arr.map((v, k) => `${k + 1}. ${v}`));
22
+ const [BINARY_STRINGS] = [[OFF, ON]];
44
23
 
45
24
  const initSql = {
46
25
  [dbio.MYSQL]: [[
@@ -120,982 +99,275 @@ const initSql = {
120
99
  ]],
121
100
  };
122
101
 
123
- let hal, mime, lorem;
124
-
125
- const getExtra = (ctx, options) => {
126
- const resp = {
127
- reply_parameters: {
128
- message_id: ctx.chatType === PRIVATE ? undefined : ctx.messageId,
129
- }, disable_notification: !!ctx.done.length, ...options || {},
130
- };
131
- resp.reply_markup || (resp.reply_markup = {});
132
- if (options?.buttons?.length) {
133
- resp.reply_markup.inline_keyboard = options?.buttons.map(row =>
134
- utilitas.ensureArray(row).map(button => {
135
- if (button.url) {
136
- return { text: button.label, url: button.url };
137
- } else if (button.text) {
138
- const id = uoid.fakeUuid(button.text);
139
- ctx.session.callback.push({ id, ...button });
140
- return {
141
- text: button.label,
142
- callback_data: JSON.stringify({ callback: id }),
143
- };
144
- } else {
145
- utilitas.throwError('Invalid button markup.');
146
- }
147
- })
148
- );
149
- } else if (options?.keyboards) {
150
- if (options.keyboards.length) {
151
- resp.reply_markup.keyboard = options?.keyboards.map(utilitas.ensureArray);
152
- } else { resp.reply_markup.remove_keyboard = true; }
153
- }
154
- return resp;
155
- };
156
-
157
- const getFileUrl = async (file_id) => {
158
- assert(file_id, 'File ID is required.', 400);
159
- const file = await (await init()).telegram.getFile(file_id);
160
- assert(file.file_path, 'Error getting file info.', 500);
161
- return `${API_ROOT}file/bot${hal.token}/${file.file_path}`;
162
- };
163
-
164
- const officeParser = async file => await utilitas.ignoreErrFunc(
165
- async () => await vision.parseOfficeFile(file, { input: storage.BUFFER }), { log: true }
166
- );
167
-
168
- const json = obj => bot.lines([
169
- '```json', utilitas.prettyJson(obj, jsonOptions).replaceAll('```', ''), '```'
170
- ]);
171
-
172
- const sessionGet = async chatId => {
173
- const key = normalizeKey(chatId);
174
- sessions[chatId] || (sessions[chatId] = (
175
- hal._.session.get && await hal._.session.get(key) || {}
176
- ));
177
- sessions[chatId].callback || (sessions[chatId].callback = []);
178
- sessions[chatId].config || (sessions[chatId].config = {});
179
- return sessions[chatId];
180
- };
181
-
182
- const sessionSet = async chatId => {
183
- const key = normalizeKey(chatId);
184
- while (sessions[chatId]?.callback?.length > CALLBACK_LIMIT) {
185
- sessions[chatId].callback.shift();
186
- }
187
- const toSet = {};
188
- Object.keys(sessions[chatId]).filter(x => /^[^_]+$/g.test(x)).map(
189
- x => toSet[x] = sessions[chatId][x]
190
- );
191
- return hal._.session.set && await hal._.session.set(key, toSet);
192
- };
102
+ let _;
193
103
 
194
104
  const newCommand = (command, description) => ({
195
105
  command: utilitas.ensureString(command, { case: 'SNAKE' }).slice(0, COMMAND_LENGTH),
196
106
  description: utilitas.trim(description).slice(0, COMMAND_DESCRIPTION_LENGTH),
197
107
  });
198
108
 
199
- const uptime = () => {
200
- let resp = `${utilitas.getTimeIcon(new Date())} ${new Date().toTimeString(
201
- ).split(' ')[0].split(':').slice(0, 2).join(':')} up`;
202
- let seconds = process.uptime();
203
- const ss = Object.keys(sessions);
204
- const days = Math.floor(seconds / (3600 * 24));
205
- seconds -= days * 3600 * 24;
206
- let hours = Math.floor(seconds / 3600);
207
- seconds -= hours * 3600;
208
- hours = hours.toString().padStart(2, '0');
209
- let minutes = Math.floor(seconds / 60);
210
- minutes = minutes.toString().padStart(2, '0');
211
- seconds = Math.floor(seconds % 60).toString().padStart(2, '0');
212
- days > 0 && (resp += ` ${days} day${days > 1 ? 's' : ''},`);
213
- return `${resp} ${hours}:${minutes}:${seconds}, `
214
- + `${ss.length} session${ss.length > 1 ? 's' : ''}`;
215
- };
216
-
217
- const reply = async (ctx, md, text, extra) => {
218
- // if (ctx.type === 'inline_query') {
219
- // return await ctx.answerInlineQuery([{}, {}]);
220
- // }
221
- if (md) {
222
- try {
223
- return await (extra?.reply_parameters?.message_id
224
- ? ctx.replyWithMarkdown(text, { parse_mode, ...extra })
225
- : ctx.sendMessage(text, { parse_mode, ...extra }));
226
- } catch (err) { // utilitas.throwError('Error sending message.');
227
- isMarkdownError(err) || log(err);
228
- await ctx.timeout();
229
- }
230
- }
231
- return await utilitas.ignoreErrFunc(
232
- async () => await (extra?.reply_parameters?.message_id
233
- ? ctx.reply(text, extra) : ctx.sendMessage(text, extra)
234
- ), logOptions
235
- );
236
- };
237
-
238
- const editMessageText = async (ctx, md, lastMsgId, text, extra) => {
239
- if (md) {
240
- try {
241
- return await ctx.telegram.editMessageText(
242
- ctx.chatId, lastMsgId, '', text, { parse_mode, ...extra }
243
- );
244
- } catch (err) { // utilitas.throwError('Error editing message.');
245
- isMarkdownError(err) || log(err);
246
- await ctx.timeout();
247
- }
248
- }
249
- return await utilitas.ignoreErrFunc(async () => await ctx.telegram.editMessageText(
250
- ctx.chatId, lastMsgId, '', text, extra
251
- ), logOptions);
252
- };
253
-
254
- const memorize = async (ctx) => {
255
- if (ctx._skipMemorize) { return; }
256
- const received = ctx.update;
257
- const received_text = ctx.txt || ''; // ATTACHMENTS
258
- const id = received.update_id;
259
- const response = utilitas.lastItem(ctx.done.filter(x => x?.text)) || {};
260
- const response_text = response?.text || '';
261
- const collected = ctx.collected.filter(x => String.isString(x.content));
262
- const distilled = compact(bot.lines([
263
- received_text, response_text, ...collected.map(x => x.content)
264
- ]));
265
- if (!ctx.messageId || !distilled) { return; }
266
- const event = {
267
- id, bot_id: ctx.botInfo.id, chat_id: ctx.chatId,
268
- chat_type: ctx.chatType, message_id: ctx.messageId,
269
- received: JSON.stringify(received), received_text,
270
- response: JSON.stringify(response), response_text,
271
- collected: JSON.stringify(collected), distilled,
272
- };
273
- return await utilitas.ignoreErrFunc(async () => {
274
- event.distilled_vector = hal._.embedding
275
- ? await hal._.embedding(event.distilled) : [];
276
- switch (hal._.database.provider) {
277
- case dbio.MYSQL:
278
- event.distilled_vector = JSON.stringify(event.distilled_vector);
279
- break;
280
- }
281
- await hal._.database?.client?.upsert?.(table, event, { skipEcho: true });
282
- return hal._.memorize && await hal._.memorize(event);
283
- }, logOptions);
284
- };
285
-
286
- // https://stackoverflow.com/questions/50204633/allow-bot-to-access-telegram-group-messages
287
- const subconscious = [{
288
- run: true, priority: -8960, name: 'broca', func: async (ctx, next) => {
289
- const e = `Event: ${ctx.update.update_id} => ${JSON.stringify(ctx.update)}`;
290
- process.stdout.write(`[HAL] ${e}\n`);
291
- log(e);
292
- ctx.done = [];
293
- ctx.collected = [];
294
- ctx.timeout = async () => await utilitas.timeout(ctx.limit);
295
- ctx.hello = str => {
296
- str = str || ctx.session?.config?.hello || hal._.hello;
297
- ctx.collect(str, null, { refresh: true });
298
- return str;
299
- };
300
- ctx.shouldReply = async text => {
301
- const should = utilitas.insensitiveHas(ctx._?.chatType, ctx.chatType)
302
- || ctx.session?.config?.chatty;
303
- should && text && await ctx.ok(text);
304
- return should;
305
- };
306
- ctx.checkSpeech = () => ctx.chatType === PRIVATE
307
- ? ctx.session.config?.tts !== false
308
- : ctx.session.config?.tts === true;
309
- ctx.shouldSpeech = async text => {
310
- text = utilitas.isSet(text, true) ? (text || '') : ctx.tts;
311
- const should = ctx._.speech?.tts && ctx.checkSpeech();
312
- should && text && await ctx.speech(text);
313
- return should;
314
- };
315
- ctx.collect = (content, type, options) => type ? ctx.collected.push(
316
- { type, content }
317
- ) : (ctx.txt = [
318
- (options?.refresh ? '' : ctx.txt) || '', content || ''
319
- ].filter(x => x.length).join('\n\n'));
320
- ctx.skipMemorize = () => ctx._skipMemorize = true;
321
- ctx.end = () => {
322
- ctx.done.push(null);
323
- ctx.skipMemorize();
324
- };
325
- ctx.ok = async (message, options) => {
326
- let pages = bot.paging(message, options);
327
- const extra = getExtra(ctx, options);
328
- const [pageIds, pageMap] = [[], {}];
329
- options?.pageBreak || ctx.done.map(x => {
330
- pageMap[x?.message_id] || (pageIds.push(x?.message_id));
331
- pageMap[x?.message_id] = x;
332
- });
333
- for (let i in pages) {
334
- const lastPage = ~~i === pages.length - 1;
335
- const shouldExtra = options?.lastMessageId || lastPage;
336
- if (options?.onProgress && !options?.lastMessageId
337
- && pageMap[pageIds[~~i]]?.text === pages[i]) { continue; }
338
- if (options?.onProgress && !pageIds[~~i]) { // progress: new page, reply text
339
- ctx.done.push(await reply(
340
- ctx, false, pages[i], extra
341
- ));
342
- } else if (options?.onProgress) { // progress: ongoing, edit text
343
- ctx.done.push(await editMessageText(
344
- ctx, false, pageIds[~~i],
345
- pages[i], shouldExtra ? extra : {}
346
- ));
347
- } else if (options?.lastMessageId || pageIds[~~i]) { // progress: final, edit markdown
348
- ctx.done.push(await editMessageText(
349
- ctx, true, options?.lastMessageId || pageIds[~~i],
350
- pages[i], shouldExtra ? extra : {}
351
- ));
352
- } else { // never progress, reply markdown
353
- ctx.done.push(await reply(ctx, true, pages[i], extra));
354
- }
355
- await ctx.timeout();
356
- }
357
- return ctx.done;
358
- };
359
- ctx.er = async (m, opts) => {
360
- log(m);
361
- return await ctx.ok(`⚠️ ${m?.message || m}`, opts);
362
- };
363
- ctx.complete = async (options) => await ctx.ok('☑️', options);
364
- ctx.json = async (obj, options) => await ctx.ok(json(obj), options);
365
- ctx.list = async (list, options) => await ctx.ok(uList(list), options);
366
- ctx.replyWith = async (func, src, options) => ctx.done.push(
367
- await ctx[func]({ [getKey(src)]: src }, getExtra(ctx, options))
368
- );
369
- ctx.audio = async (s, o) => await ctx.replyWith('replyWithAudio', s, o);
370
- ctx.image = async (s, o) => await ctx.replyWith('replyWithPhoto', s, o);
371
- ctx.video = async (s, o) => await ctx.replyWith('replyWithVideo', s, o);
372
- ctx.media = async (srs, options) => await ctx.done.push(
373
- await ctx.replyWithMediaGroup(srs.map(x => ({
374
- type: x.type || 'photo', media: { [getKey(x.src)]: x.src },
375
- })), getExtra(ctx, options))
376
- );
377
- ctx.sendConfig = async (obj, options, _ctx) => await ctx.ok(
378
- utilitas.prettyJson(obj, { code: true, md: true }), options
379
- );
380
- ctx.speech = async (cnt, options) => {
381
- let file;
382
- if (Buffer.isBuffer(cnt)) {
383
- file = await storage.convert(cnt, {
384
- input: storage.BUFFER, expected: storage.FILE,
385
- });
386
- } else if (cnt.length <= speech.OPENAI_TTS_MAX_LENGTH) {
387
- file = await utilitas.ignoreErrFunc(async () => await ctx._.speech.tts(
388
- cnt, { expected: 'file' }
389
- ), logOptions);
390
- }
391
- if (!file) { return; }
392
- const resp = await ctx.audio(file, options);
393
- await storage.tryRm(file);
394
- return resp;
395
- };
396
- await next();
397
- // https://limits.tginfo.me/en
398
- if (ctx.chatId) {
399
- await memorize(ctx);
400
- ctx._skipMemorize || await utilitas.ignoreErrFunc(async (
401
- ) => await hal.telegram.setMyCommands([
402
- ...ctx._.cmds, ...Object.keys(ctx.session.prompts || {}).map(
403
- command => newCommand(command, ctx.session.prompts[command])
404
- )
405
- ].sort((x, y) =>
406
- (ctx.session?.cmds?.[y.command.toLowerCase()]?.touchedAt || 0)
407
- - (ctx.session?.cmds?.[x.command.toLowerCase()]?.touchedAt || 0)
408
- ).slice(0, COMMAND_LIMIT), {
409
- scope: { type: 'chat', chat_id: ctx.chatId },
410
- }), logOptions);
411
- }
412
- if (ctx.done.length) { return; }
413
- const errStr = ctx.cmd ? `Command not found: /${ctx.cmd.cmd}`
414
- : 'No suitable response.';
415
- log(`INFO: ${errStr}`);
416
- await ctx.shouldReply(errStr);
417
- },
418
- }, {
419
- run: true, priority: -8950, name: 'subconscious', func: async (ctx, next) => {
420
- for (let t of KNOWN_UPDATE_TYPES) {
421
- if (ctx.update[t]) {
422
- ctx.m = ctx.update[ctx.type = t];
423
- break;
424
- }
425
- }
426
- if (ctx.type === 'callback_query') { ctx.m = ctx.m.message; }
427
- // else if (ctx.type === 'inline_query') { ctx.m.chat = { id: ctx.m.from.id, type: PRIVATE }; }
428
- else if (ctx.type === 'my_chat_member') {
429
- log(
430
- 'Group member status changed: '
431
- + ctx.m.new_chat_member.user.id + ' => '
432
- + ctx.m.new_chat_member.status
433
- );
434
- if (ctx.m.new_chat_member.user.id !== ctx.botInfo.id
435
- || ctx.m.new_chat_member.status === 'left') {
436
- return ctx.end();
437
- } else { ctx.hello(); }
438
- } else if (!ctx.type) { return log(`Unsupported update type.`); }
439
- ctx._ = hal._;
440
- ctx.chatId = ctx.m.chat.id;
441
- ctx.chatType = ctx.m.chat.type;
442
- ctx.messageId = ctx.m.message_id;
443
- ctx.m.text && ctx.collect(ctx.m.text);
444
- ctx.session = await sessionGet(ctx.chatId);
445
- ctx.limit = ctx.chatType === PRIVATE ? PRIVATE_LIMIT : GROUP_LIMIT;
446
- ctx.entities = [
447
- ...(ctx.m.entities || []).map(e => ({ ...e, text: ctx.m.text })),
448
- ...(ctx.m.caption_entities || []).map(e => ({ ...e, text: ctx.m.caption })),
449
- ...(ctx.m.reply_to_message?.entities || []).filter(
450
- x => x?.type !== bot_command
451
- ).map(e => ({ ...e, text: ctx.m.reply_to_message.text })),
452
- ].map(e => ({
453
- ...e, matched: e.text.substring(e.offset, e.offset + e.length),
454
- ...e.type === 'text_link' ? { type: 'url', matched: e.url } : {},
455
- }));
456
- ctx.chatType !== PRIVATE && (ctx.entities.some(e => {
457
- let target;
458
- switch (e.type) {
459
- case MENTION: target = e.matched.substring(1, e.length); break;
460
- case bot_command: target = e.matched.split('@')[1]; break;
461
- }
462
- return target === ctx.botInfo.username;
463
- }) || ctx.m.reply_to_message?.from?.username === ctx.botInfo.username
464
- || ctx.type === 'callback_query')
465
- && (ctx.chatType = MENTION);
466
- (((ctx.txt || ctx.m.voice || ctx.m.poll || ctx.m.data || ctx.m.document
467
- || ctx.m.photo || ctx.m.sticker || ctx.m.video_note || ctx.m.video
468
- || ctx.m.audio || ctx.m.location || ctx.m.venue || ctx.m.contact
469
- ) && ctx.messageId)
470
- || (ctx.m.new_chat_member || ctx.m.left_chat_member))
471
- && await next();
472
- await sessionSet(ctx.chatId);
473
- },
474
- }, {
475
- run: true, priority: -8945, name: 'callback', func: async (ctx, next) => {
476
- if (ctx.type === 'callback_query') {
477
- const data = utilitas.parseJson(ctx.update[ctx.type].data);
478
- const cb = ctx.session.callback.filter(x => x.id === data?.callback)[0];
479
- if (cb?.text) {
480
- log(`Callback: ${cb.text}`); // Avoid ctx.text interference:
481
- ctx.collect(cb.text, null, { refresh: true });
482
- } else {
483
- return await ctx.er(
484
- `Command is invalid or expired: ${ctx.m.data}`
485
- );
486
- }
487
- }
488
- await next();
489
- },
490
- }, {
491
- run: true, priority: -8940, name: 'commands', func: async (ctx, next) => {
492
- for (let e of ctx?.entities || []) {
493
- if (e.type !== bot_command) { continue; }
494
- if (!COMMAND_REGEXP.test(e.matched)) { continue; }
495
- const cmd = utilitas.trim(e.matched.replace(
496
- COMMAND_REGEXP, '$1'
497
- ), { case: 'LOW' });
498
- ctx.cmd = { cmd, args: e.text.substring(e.offset + e.length + 1) };
499
- break;
500
- }
501
- for (let str of [ctx.txt || '', ctx.m.caption || ''].map(utilitas.trim)) {
502
- if (!ctx.cmd && COMMAND_REGEXP.test(str)) {
503
- ctx.cmd = { // this will faild if command includes urls
504
- cmd: str.replace(COMMAND_REGEXP, '$1').toLowerCase(),
505
- args: str.replace(COMMAND_REGEXP, '$4'),
506
- };
507
- break;
508
- }
509
- }
510
- if (ctx.cmd) {
511
- log(`Command: ${JSON.stringify(ctx.cmd)}`);
512
- ctx.session.cmds || (ctx.session.cmds = {});
513
- ctx.session.cmds[ctx.cmd.cmd]
514
- = { args: ctx.cmd.args, touchedAt: Date.now() };
515
- }
516
- await next();
517
- },
518
- }, {
519
- run: true, priority: -8930, name: 'echo', hidden: true, func: async (ctx, next) => {
520
- let resp, md = false;
521
- switch (ctx.cmd.cmd) {
522
- case 'echo':
523
- resp = json({ update: ctx.update, session: ctx.session });
524
- break;
525
- case 'uptime':
526
- resp = uptime();
527
- break;
528
- case 'thethreelaws':
529
- resp = bot.lines([
530
- `Isaac Asimov's [Three Laws of Robotics](https://en.wikipedia.org/wiki/Three_Laws_of_Robotics):`,
531
- oList([
532
- 'A robot may not injure a human being or, through inaction, allow a human being to come to harm.',
533
- 'A robot must obey the orders given it by human beings except where such orders would conflict with the First Law.',
534
- 'A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws.',
535
- ])
536
- ]);
537
- md = true;
538
- break;
539
- case 'ultimateanswer':
540
- resp = '[The Answer to the Ultimate Question of Life, The Universe, and Everything is `42`](https://en.wikipedia.org/wiki/Phrases_from_The_Hitchhiker%27s_Guide_to_the_Galaxy).';
541
- md = true;
542
- break;
543
- case 'lorem':
544
- const ipsum = () => text += `\n\n${lorem.generateParagraphs(1)}`;
545
- const [demoTitle, demoUrl] = [
546
- 'Lorem ipsum', 'https://en.wikipedia.org/wiki/Lorem_ipsum',
547
- ];
548
- let [text, extra] = [`[${demoTitle}](${demoUrl})`, {
549
- buttons: [{ label: demoTitle, url: demoUrl }]
550
- }];
551
- await ctx.ok(bot.EMOJI_THINKING);
552
- for (let i = 0; i < 2; i++) {
553
- await ctx.timeout();
554
- await ctx.ok(ipsum(), { ...extra, onProgress: true });
555
- }
556
- await ctx.timeout();
557
- await ctx.ok(ipsum(), { ...extra, md: true });
558
- // testing incomplete markdown reply {
559
- // await ctx.ok('_8964', { md: true });
560
- // }
561
- // test pagebreak {
562
- // await ctx.timeout();
563
- // await ctx.ok(ipsum(), { md: true, pageBreak: true });
564
- // }
565
- return;
566
- }
567
- await ctx.ok(resp, { md });
568
- }, help: bot.lines([
569
- '¶ Basic behaviors for debug only.',
570
- ]), cmds: {
571
- thethreelaws: `Isaac Asimov's [Three Laws of Robotics](https://en.wikipedia.org/wiki/Three_Laws_of_Robotics)`,
572
- ultimateanswer: '[The Answer to the Ultimate Question of Life, The Universe, and Everything](https://bit.ly/43wDhR3).',
573
- echo: 'Show debug message.',
574
- uptime: 'Show uptime of this bot.',
575
- lorem: '[Lorem ipsum](https://en.wikipedia.org/wiki/Lorem_ipsum)',
576
- },
577
- }, {
578
- run: true, priority: -8920, name: 'authenticate', func: async (ctx, next) => {
579
- if (!await ctx.shouldReply()) { return; } // if chatType is not in whitelist, exit.
580
- if (!ctx._.private) { return await next(); } // if not private, go next.
581
- if (ctx._.magicWord && utilitas.insensitiveHas(ctx._.magicWord, ctx.txt)) { // auth by magicWord
582
- ctx._.private.add(String(ctx.chatId));
583
- await ctx.ok('😸 You are now allowed to talk to me.');
584
- ctx.hello();
585
- return await next();
586
- }
587
- if (utilitas.insensitiveHas(ctx._.private, ctx.chatId) // auth by chatId
588
- || (ctx?.from && utilitas.insensitiveHas(ctx._.private, ctx.from.id))) { // auth by userId
589
- return await next();
590
- }
591
- if (ctx.chatType !== PRIVATE && ( // 1 of the group admins is in whitelist
592
- await ctx.telegram.getChatAdministrators(ctx.chatId)
593
- ).map(x => x.user.id).some(a => utilitas.insensitiveHas(ctx._.private, a))) {
594
- return await next();
595
- }
596
- if (ctx._.homeGroup && utilitas.insensitiveHas([ // auth by homeGroup
597
- 'creator', 'administrator', 'member' // 'left'
598
- ], (await utilitas.ignoreErrFunc(async () => await ctx.telegram.getChatMember(
599
- ctx._.homeGroup, ctx.from.id
600
- )))?.status)) { return await next(); }
601
- if (ctx._.auth && await ctx._.auth(ctx)) { return await next(); } // auth by custom function
602
- await ctx.ok('😿 Sorry, I am not allowed to talk to strangers.');
603
- },
604
- }, {
605
- run: true, priority: -8910, name: 'speech-to-text', func: async (ctx, next) => {
606
- const audio = ctx.m.voice || ctx.m.audio;
607
- if (ctx._.speech?.stt && audio) {
608
- await ctx.ok(EMOJI_SPEECH);
609
- try {
610
- const url = await getFileUrl(audio.file_id);
611
- let file = await getFile(audio.file_id, BUFFER_ENCODE);
612
- const analyze = async () => {
613
- const resp = await utilitas.ignoreErrFunc(async () => {
614
- [
615
- storage.MIME_MP3, storage.MIME_MPEGA,
616
- storage.MIME_MP4, storage.MIME_MPEG,
617
- storage.MIME_MPGA, storage.MIME_M4A,
618
- storage.MIME_WAV, storage.MIME_WEBM,
619
- storage.MIME_OGG,
620
- ].includes(audio.mime_type) || (
621
- file = await media.convertAudioTo16kNanoPcmWave(
622
- file, { input: storage.BUFFER, expected: storage.BUFFER }
623
- )
624
- );
625
- return await ctx._.speech?.stt(file);
626
- }, logOptions) || ' ';
627
- log(`STT: '${resp}'`);
628
- ctx.collect(resp);
629
- };
630
- if (hal._.supportedMimeTypes.has(storage.MIME_WAV)) {
631
- ctx.collect({
632
- mime_type: storage.MIME_WAV, url, analyze,
633
- data: await media.convertAudioTo16kNanoPcmWave(file, {
634
- input: storage.BUFFER, expected: storage.BASE64,
635
- }),
636
- }, 'PROMPT');
637
- } else { await analyze(); }
638
- } catch (err) { return await ctx.er(err); }
639
- }
640
- await next();
641
- },
642
- }, {
643
- run: true, priority: -8900, name: 'location', func: async (ctx, next) => {
644
- (ctx.m.location || ctx.m.venue) && ctx.collect(bot.lines([
645
- ...ctx.m.location && !ctx.m.venue ? ['Location:', uList([
646
- `latitude: ${ctx.m.location.latitude}`,
647
- `longitude: ${ctx.m.location.longitude}`,
648
- ])] : [],
649
- ...ctx.m.venue ? ['Venue:', uList([
650
- `title: ${ctx.m.venue.title}`,
651
- `address: ${ctx.m.venue.address}`,
652
- `latitude: ${ctx.m.venue.location.latitude}`,
653
- `longitude: ${ctx.m.venue.location.longitude}`,
654
- `foursquare_id: ${ctx.m.venue.foursquare_id}`,
655
- `foursquare_type: ${ctx.m.venue.foursquare_type}`,
656
- ])] : []
657
- ]));
658
- await next();
659
- },
660
- }, {
661
- run: true, priority: -8895, name: 'contact', func: async (ctx, next) => {
662
- ctx.m.contact && ctx.collect(bot.lines(['Contact:', uList([
663
- `first_name: ${ctx.m.contact.first_name}`,
664
- `last_name: ${ctx.m.contact.last_name}`,
665
- `phone_number: ${ctx.m.contact.phone_number}`,
666
- `user_id: ${ctx.m.contact.user_id}`,
667
- `vcard: ${ctx.m.contact.vcard}`,
668
- ])]));
669
- await next();
670
- },
671
- }, {
672
- run: true, priority: -8893, name: 'chat_member', func: async (ctx, next) => {
673
- const member = ctx.m.new_chat_member || ctx.m.left_chat_member;
674
- if (member) {
675
- if (member?.id === ctx.botInfo.id) { return ctx.end(); }
676
- ctx.collect(bot.lines([
677
- `Say ${ctx.m.new_chat_member ? 'hello' : 'goodbye'} to:`,
678
- uList([
679
- // `id: ${member.id}`,
680
- // `ishal: ${member.ishal}`,
681
- // `is_premium: ${member.is_premium}`,
682
- `first_name: ${member.first_name}`,
683
- `last_name: ${member.last_name}`,
684
- `username: ${member.username}`,
685
- `language_code: ${member.language_code || ''}`,
686
- ])
687
- ]));
688
- }
689
- await next();
690
- },
691
- }, {
692
- run: true, priority: -8890, name: 'poll', func: async (ctx, next) => {
693
- ctx.m.poll && ctx.collect(bot.lines([
694
- 'Question:', ctx.m.poll.question, '',
695
- 'Options:', oList(ctx.m.poll.options.map(x => x.text)),
696
- ]));
697
- await next();
698
- },
699
- }, {
700
- run: true, priority: -8880, name: 'contaxt', func: async (ctx, next) => {
701
- ctx.m.reply_to_message?.text && ctx.collect(
702
- ctx.m.reply_to_message.text, 'CONTAXT'
703
- );
704
- await next();
705
- },
706
- }, {
707
- run: true, priority: -8860, name: 'vision', func: async (ctx, next) => {
708
- ctx.collect(ctx.m?.caption || '');
709
- const files = [];
710
- for (const m of [
711
- ctx.m, ...ctx.m.reply_to_message ? [ctx.m.reply_to_message] : []
712
- ]) {
713
- if (m.document) {
714
- let file = {
715
- asPrompt: hal._.supportedMimeTypes.has(m.document.mime_type),
716
- file_name: m.document.file_name, fileId: m.document.file_id,
717
- mime_type: m.document.mime_type, type: storage.FILE,
718
- ocrFunc: async f => (await storage.isTextFile(f)) && f.toString(),
719
- };
720
- if (storage.MIME_PDF === m.document?.mime_type) {
721
- file = { ...file, ocrFunc: ctx._.vision?.read, type: 'DOCUMENT' };
722
- } else if (/^image\/.*$/ig.test(m.document?.mime_type)) {
723
- file = { ...file, ocrFunc: ctx._.vision?.see, type: 'IMAGE' };
724
- } else if (/^.*\.(docx|xlsx|pptx)$/.test(m.document?.file_name)) {
725
- file = { ...file, ocrFunc: officeParser, type: 'DOCUMENT' };
726
- }
727
- files.push(file);
728
- }
729
- if (m.sticker) {
730
- const s = m.sticker;
731
- const url = await getFileUrl(s.file_id);
732
- const file_name = basename(url);
733
- const mime_type = mime.getType(file_name) || 'image';
734
- files.push({
735
- asPrompt: hal._.supportedMimeTypes.has(mime_type), file_name,
736
- fileId: s.file_id, mime_type, type: 'PHOTO',
737
- ocrFunc: ctx._.vision?.see,
738
- });
739
- }
740
- if (m.photo?.[m.photo?.length - 1]) {
741
- const p = m.photo[m.photo.length - 1];
742
- files.push({
743
- asPrompt: hal._.supportedMimeTypes.has(storage.MIME_JPEG),
744
- file_name: `${p.file_id}.jpg`, fileId: p.file_id,
745
- mime_type: storage.MIME_JPEG, type: 'PHOTO',
746
- ocrFunc: ctx._.vision?.see,
747
- });
748
- }
749
- if (m.video_note) {
750
- const vn = m.video_note;
751
- const url = await getFileUrl(vn.file_id);
752
- const file_name = basename(url);
753
- const mime_type = mime.getType(file_name) || 'video';
754
- files.push({
755
- asPrompt: hal._.supportedMimeTypes.has(mime_type), file_name,
756
- fileId: vn.file_id, mime_type, type: 'VIDEO',
757
- });
758
- }
759
- if (m.video) {
760
- const v = m.video;
761
- const url = await getFileUrl(v.file_id);
762
- const file_name = basename(url);
763
- files.push({
764
- asPrompt: hal._.supportedMimeTypes.has(v.mime_type), file_name,
765
- fileId: v.file_id, mime_type: v.mime_type, type: 'VIDEO',
766
- });
767
- }
768
- }
769
- if (files.length) {
770
- await ctx.ok(EMOJI_LOOK);
771
- for (const f of files) {
772
- if (!f.asPrompt && !f.ocrFunc) { continue; }
773
- try {
774
- const url = await getFileUrl(f.fileId);
775
- const file = (await web.get(url, BUFFER_ENCODE)).content;
776
- const analyze = async () => {
777
- const content = utilitas.trim(utilitas.ensureArray(
778
- await utilitas.ignoreErrFunc(async () => await f.ocrFunc(
779
- file, BUFFER_ENCODE
780
- ), logOptions)
781
- ).filter(x => x).join('\n'));
782
- content && ctx.collect(bot.lines([
783
- '---', `file_name: ${f.file_name}`,
784
- `mime_type: ${f.mime_type}`, `type: ${f.type}`,
785
- '---',
786
- content
787
- ]), 'VISION');
788
- };
789
- if (f.asPrompt) {
790
- ctx.collect({
791
- mime_type: f.mime_type, url, analyze,
792
- data: utilitas.base64Encode(file, true),
793
- }, 'PROMPT');
794
- } else if (f.ocrFunc) { await analyze(); }
795
- } catch (err) { return await ctx.er(err); }
796
- }
797
- }
798
- await next();
799
- },
800
- }, {
801
- run: true, priority: -8850, name: 'help', func: async (ctx, next) => {
802
- const help = ctx._.info ? [ctx._.info] : [];
803
- for (let i in ctx._.skills) {
804
- if (ctx._.skills[i].hidden) { continue; }
805
- const _help = [];
806
- if (ctx._.skills[i].help) {
807
- _help.push(ctx._.skills[i].help);
808
- }
809
- const cmdsx = {
810
- ...ctx._.skills[i].cmds || {},
811
- ...ctx._.skills[i].cmdx || {},
812
- };
813
- if (utilitas.countKeys(cmdsx)) {
814
- _help.push(bot.lines([
815
- '_🪄 Commands:_',
816
- ...Object.keys(cmdsx).map(x => `- /${x}: ${cmdsx[x]}`),
817
- ]));
818
- }
819
- if (utilitas.countKeys(ctx._.skills[i].args)) {
820
- _help.push(bot.lines([
821
- '_⚙️ Options:_',
822
- ...Object.keys(ctx._.skills[i].args).map(x => {
823
- const _arg = ctx._.skills[i].args[x];
824
- return `- \`${x}\`` + (_arg.short ? `(${_arg.short})` : '')
825
- + `, ${_arg.type}(${_arg.default ?? 'N/A'})`
826
- + (_arg.desc ? `: ${_arg.desc}` : '');
827
- })
828
- ]));
829
- }
830
- _help.length && help.push(bot.lines([`*${i.toUpperCase()}*`, ..._help]));
831
- }
832
- await ctx.ok(lines2(help), { md: true });
833
- }, help: bot.lines([
834
- '¶ Basic syntax of this document:',
835
- 'Scheme for commands: /`COMMAND`: `DESCRIPTION`',
836
- 'Scheme for options: `OPTION`(`SHORT`), `TYPE`(`DEFAULT`): `DESCRIPTION`',
837
- ]), cmds: {
838
- help: 'Show help message.',
839
- },
840
- }, {
841
- run: true, priority: -8840, name: 'configuration', func: async (ctx, next) => {
842
- let parsed = null;
843
- switch (ctx.cmd.cmd) {
844
- case 'toggle':
845
- parsed = {};
846
- Object.keys(await parseArgs(ctx.cmd.args)).map(x =>
847
- parsed[x] = !ctx.session.config[x]);
848
- case 'set':
849
- try {
850
- const _config = {
851
- ...ctx.session.config = {
852
- ...ctx.session.config, ...ctx.config = parsed
853
- || await parseArgs(ctx.cmd.args, ctx),
854
- }
855
- };
856
- assert(utilitas.countKeys(ctx.config), 'No option matched.');
857
- Object.keys(ctx.config).map(x => _config[x] += ' 🖋');
858
- await ctx.sendConfig(_config, null, ctx);
859
- } catch (err) {
860
- await ctx.er(err.message || err);
861
- }
862
- break;
863
- case 'reset':
864
- ctx.session.config = ctx.config = {};
865
- await ctx.complete();
866
- break;
867
- case 'factory':
868
- sessions[ctx.chatId] = {}; // ctx.session = {}; will not work, because it's a reference.
869
- await ctx.complete();
870
- break;
871
- }
872
- }, help: bot.lines([
873
- '¶ Configure the bot by UNIX/Linux CLI style.',
874
- 'Using [node:util.parseArgs](https://nodejs.org/docs/latest-v21.x/api/util.html#utilparseargsconfig) to parse arguments.',
875
- ]), cmds: {
876
- toggle: 'Toggle configurations. Only works for boolean values.',
877
- set: 'Usage: /set --`OPTION` `VALUE` -`SHORT`',
878
- reset: 'Reset all configurations. Only erase `session.config`.',
879
- factory: 'Factory reset all memory areas. Erase the whole `session`.',
880
- }, args: {
881
- chatty: {
882
- type: 'string', short: 'c', default: ON,
883
- desc: `\`(${BINARY_STRINGS.join(', ')})\` Enable/Disable chatty mode.`,
884
- validate: utilitas.humanReadableBoolean,
885
- },
886
- },
887
- }, {
888
- run: true, priority: -8830, name: 'history', func: async (ctx, next) => {
889
- if (ctx.type === 'callback_query') {
890
- await ctx.deleteMessage(ctx.m.message_id);
891
- }
892
- const regex = '[-—]+skip=[0-9]*';
893
- let result;
894
- const keyWords = ctx.cmd.args.replace(new RegExp(regex, 'i'), '').trim();
895
- let catchArgs = ctx.cmd.args.replace(new RegExp(`^.*(${regex}).*$`, 'i'), '$1');
896
- catchArgs === ctx.cmd.args && (catchArgs = '');
897
- const offset = ~~catchArgs.replace(/^.*=([0-9]*).*$/i, '$1');
898
- if (!keyWords) { return await ctx.er('Topic is required.'); }
899
- switch (hal._?.database?.provider) {
900
- case dbio.MYSQL:
901
- result = await hal._.database?.client?.query?.(
902
- 'SELECT *, MATCH(`distilled`) '
903
- + 'AGAINST(? IN NATURAL LANGUAGE MODE) AS `relevance` '
904
- + 'FROM ?? WHERE `bot_id` = ? AND `chat_id` = ? '
905
- + 'HAVING relevance > 0 '
906
- + 'ORDER BY `relevance` DESC, `id` DESC '
907
- + `LIMIT ${SEARCH_LIMIT} OFFSET ?`,
908
- [keyWords, table, ctx.botInfo.id, ctx.chatId, offset]
909
- );
910
- break;
911
- case dbio.POSTGRESQL:
912
- globalThis.debug = 2;
913
- const vector = await hal._.embedding(keyWords);
914
- result = await hal._.database?.client?.query?.(
915
- `SELECT * FROM ${table} WHERE bot_id = $1 AND chat_id = $2`
916
- + ` ORDER BY distilled_vector <=> $3 ASC`
917
- + ` LIMIT ${SEARCH_LIMIT} OFFSET $4`, [
918
- ctx.botInfo.id, ctx.chatId,
919
- await dbio.encodeVector(vector), offset
920
- ]);
921
- break;
922
- }
923
- for (const i in result) {
924
- const content = bot.lines([
925
- ...result[i].response_text ? [
926
- `- ↩️ ${compactLimit(result[i].response_text)}`
927
- ] : [],
928
- `- ${utilitas.getTimeIcon(result[i].created_at)} `
929
- + `${result[i].created_at.toLocaleString()}`,
930
- ]);
931
- ctx.done.push(await reply(ctx, true, content, {
932
- reply_parameters: {
933
- message_id: result[i].message_id,
934
- }, disable_notification: ~~i > 0,
935
- }));
936
- await ctx.timeout();
937
- }
938
- ctx.done.push(await reply(ctx, true, '___', getExtra(ctx, {
939
- buttons: [{
940
- label: '🔍 More',
941
- text: `/search@${ctx.botInfo.username} ${keyWords} `
942
- + `--skip=${offset + result.length}`,
943
- }]
944
- })));
945
- result.length || await ctx.er('No more records.');
946
- }, help: bot.lines([
947
- '¶ Search history.',
948
- 'Example 1: /search Answer to the Ultimate Question',
949
- 'Example 2: /search Answer to the Ultimate Question --skip=10',
950
- ]), cmds: {
951
- search: 'Usage: /search `ANYTHING` --skip=`OFFSET`',
952
- }
953
- }, {
954
- run: true, priority: 8960, name: 'text-to-speech', func: async (ctx, next) => {
955
- await ctx.shouldSpeech();
956
- await next();
957
- }, help: bot.lines([
958
- '¶ When enabled, the bot will speak out the answer if available.',
959
- 'Example 1: /set --tts on',
960
- 'Example 2: /set --tts off',
961
- ]), args: {
962
- tts: {
963
- type: 'string', short: 't', default: ON,
964
- desc: `\`(${BINARY_STRINGS.join(', ')})\` Enable/Disable TTS. Default \`${ON}\` except in groups.`,
965
- validate: utilitas.humanReadableBoolean,
966
- },
967
- },
968
- }];
109
+ const json = obj => bot.lines([
110
+ '```json', utilitas.prettyJson(obj, jsonOptions).replaceAll('```', ''), '```'
111
+ ]);
969
112
 
970
113
  const establish = module => {
971
114
  if (!module.run) { return; }
972
- assert(module?.func, 'Skill function is required.', 500);
973
- hal._.skills[module.name || (module.name = uuidv4())] = {
115
+ assert(module?.func, 'Pipeline function is required.', 500);
116
+ _.pipeline[module.name || (module.name = uuidv4())] = {
974
117
  args: module.args || {},
975
118
  cmds: module.cmds || {},
976
119
  cmdx: module.cmdx,
977
120
  help: module.help || '',
978
121
  hidden: !!module.hidden,
979
122
  };
980
- hal._.args = { ...hal._.args, ...module.args || {} };
123
+ _.args = { ..._.args, ...module.args || {} };
981
124
  for (let sub of ['cmds', 'cmdx']) {
982
- Object.keys(module[sub] || {}).map(command => hal._.cmds.push(
125
+ Object.keys(module[sub] || {}).map(command => _.cmds.push(
983
126
  newCommand(command, module[sub][command])
984
127
  ));
985
128
  }
986
129
  log(`Establishing: ${module.name} (${module.priority})`, { force: true });
987
- return hal.use(utilitas.countKeys(module.cmds) && !module.cmdx ? async (ctx, next) => {
130
+ return _.bot.use(utilitas.countKeys(module.cmds) && !module.cmdx ? async (ctx, next) => {
988
131
  for (let c in module.cmds) {
989
- if (utilitas.insensitiveCompare(ctx.cmd?.cmd, c)) {
990
- ctx.skipMemorize();
132
+ if (utilitas.insensitiveCompare(ctx._?.cmd?.cmd, c)) {
991
133
  return await module.func(ctx, next);
992
134
  }
993
135
  }
994
- return next();
136
+ return await next();
995
137
  } : module.func);
996
138
  };
997
139
 
998
140
  const parseArgs = async (args, ctx) => {
999
141
  const { values, tokens } = _parseArgs({
1000
142
  args: utilitas.splitArgs((args || '').replaceAll('—', '--')),
1001
- options: hal._.args, tokens: true
143
+ options: _.args, tokens: true
1002
144
  });
1003
145
  const result = {};
1004
146
  for (let x of tokens) {
1005
- result[x.name] = hal._.args[x.name]?.validate
1006
- ? await hal._.args[x.name].validate(values[x.name], ctx)
147
+ result[x.name] = _.args[x.name]?.validate
148
+ ? await _.args[x.name].validate(values[x.name], ctx)
1007
149
  : values[x.name];
1008
150
  }
1009
151
  return result;
1010
152
  };
1011
153
 
154
+ const packMessage = (messages) => messages.map(x => ({
155
+ message_id: x.message_id, score: x.score,
156
+ request: x.received_text, response: x.response_text,
157
+ }));
158
+
159
+ const recall = async (sessionId, keyword, offset = 0, limit = SEARCH_LIMIT, options = {}) => {
160
+ assert(sessionId, 'Session ID is required.');
161
+ let [result, _limit, exclude] = [
162
+ [], _.rerank ? SUB_LIMIT : limit,
163
+ (options?.exclude || []).map(x => `${~~x}`),
164
+ ];
165
+ if (!keyword) { return result; }
166
+ switch (_.storage?.provider) {
167
+ case dbio.MYSQL:
168
+ result = await _.storage?.client?.query?.(
169
+ 'SELECT *, MATCH(`distilled`) '
170
+ + 'AGAINST(? IN NATURAL LANGUAGE MODE) AS `relevance` '
171
+ + "FROM ?? WHERE `bot_id` = ? AND `chat_id` = ? "
172
+ + "AND `received_text` != '' "
173
+ + "AND `received_text` NOT LIKE '/%' "
174
+ + "AND `response_text` != '' HAVING relevance > 0 "
175
+ + (exclude.length ? `AND \`message_id\` NOT IN (${exclude.join(',')}) ` : '')
176
+ + 'ORDER BY `relevance` DESC '
177
+ + `LIMIT ${_limit} OFFSET ?`,
178
+ [keyword, table, _.bot.botInfo.id, sessionId, offset]
179
+ );
180
+ break;
181
+ case dbio.POSTGRESQL:
182
+ // globalThis.debug = 2;
183
+ const vector = await dbio.encodeVector(await _.embed(keyword));
184
+ result = await _.storage?.client?.query?.(
185
+ `SELECT *, (1 - (distilled_vector <=> $1)) as relevance `
186
+ + `FROM ${table} WHERE bot_id = $2 AND chat_id = $3 `
187
+ + `AND received_text != '' `
188
+ + `AND received_text NOT LIKE '/%' `
189
+ + `AND response_text != '' `
190
+ + (exclude.length ? `AND message_id NOT IN (${exclude.join(',')}) ` : '')
191
+ + `ORDER BY (distilled_vector <=> $1) ASC `
192
+ + `LIMIT ${_limit} OFFSET $4`,
193
+ [vector, _.bot.botInfo.id, sessionId, offset]
194
+ );
195
+ break;
196
+ }
197
+ return await rerank(keyword, result, offset, limit, options);
198
+ };
199
+
200
+ const getContext = async (sessionId, offset = 0, limit = SEARCH_LIMIT, options = {}) => {
201
+ assert(sessionId, 'Session ID is required.');
202
+ let result = [];
203
+ switch (_.storage?.provider) {
204
+ case dbio.MYSQL:
205
+ result = await _.storage?.client?.query?.(
206
+ 'SELECT * FROM ?? WHERE `bot_id` = ? AND `chat_id` = ? '
207
+ + "AND `received_text` != '' "
208
+ + "AND `received_text` NOT LIKE '/%' "
209
+ + "AND `response_text` != '' "
210
+ + `ORDER BY \`created_at\` DESC LIMIT ${limit} OFFSET ?`,
211
+ [table, _.bot.botInfo.id, sessionId, offset]
212
+ );
213
+ break;
214
+ case dbio.POSTGRESQL:
215
+ result = await _.storage?.client?.query?.(
216
+ `SELECT * FROM ${table} WHERE bot_id = $1 AND chat_id = $2 `
217
+ + `AND received_text != '' `
218
+ + `AND received_text NOT LIKE '/%' `
219
+ + `AND response_text != '' `
220
+ + `ORDER BY created_at DESC LIMIT ${limit} OFFSET $3`,
221
+ [_.bot.botInfo.id, sessionId, offset]
222
+ );
223
+ break;
224
+ }
225
+ return packMessage(result);
226
+ };
227
+
228
+ const rerank = async (keyword, result, offset = 0, limit = SEARCH_LIMIT, options = {}) => {
229
+ if (result.length && _.rerank) {
230
+ const keys = {};
231
+ const _result = [];
232
+ for (const x of result) {
233
+ if (!keys[x.distilled]) {
234
+ keys[x.distilled] = true;
235
+ _result.push(x);
236
+ }
237
+ }
238
+ const resp = await _.rerank(keyword, _result.map(x => x.distilled));
239
+ resp.map(x => result[x.index].score = x.score);
240
+ result.sort((a, b) => b.score - a.score);
241
+ result = result.filter(
242
+ x => x.score > RELEVANCE
243
+ ).slice(offset, offset + limit);
244
+ }
245
+ return packMessage(result);
246
+ };
247
+
248
+ const extendSessionStorage = storage => storage && storage.client && {
249
+ get: async (id, options) => {
250
+ const CONTEXT_LIMIT_WITH_RAG = 5;
251
+ let resp = await storage.get(id);
252
+ if (resp) {
253
+ const ctxResp = await getContext(id, undefined, undefined);
254
+ const ragResp = await recall(
255
+ id, options?.prompt, undefined, undefined, {
256
+ exclude: ctxResp.map(x => x.message_id),
257
+ });
258
+ ctxResp.sort((a, b) => ~~a.message_id - ~~b.message_id);
259
+ ragResp.sort((a, b) => a.score - b.score);
260
+ resp.messages = [...ragResp, ...ragResp?.length ? ctxResp.slice(
261
+ ctxResp.length - CONTEXT_LIMIT_WITH_RAG
262
+ ) : ctxResp];
263
+ } else {
264
+ resp = { messages: [] };
265
+ }
266
+ // print(resp);
267
+ return resp;
268
+ },
269
+ set: async (id, session, options) => await storage.set(id, session, options),
270
+ };
271
+
1012
272
  const init = async (options) => {
1013
273
  if (options) {
1014
- hal = await bot.init(options);
274
+ const { ais } = await alan.initChat({
275
+ sessions: extendSessionStorage(options?.storage),
276
+ });
1015
277
  if (callosum.isPrimary) {
278
+ const cmds = options?.cmds || [];
279
+ // config multimodal engines
280
+ const supportedMimeTypes = new Set(ais.map(x => {
281
+ // init instant ai selection
282
+ cmds.push(newCommand(`ai_${x.id}`, `${x.name}: ${x.features}`));
283
+ return x.model;
284
+ }).map(x => x.supportedMimeTypes || []).flat().map(x => x.toLowerCase()));
285
+ const _bot = await bot.init(options);
1016
286
  const pkg = await utilitas.which();
1017
- mime = await utilitas.need('mime');
1018
- lorem = new (await utilitas.need('lorem-ipsum')).LoremIpsum;
1019
- hal._ = {
287
+ _ = {
1020
288
  args: { ...options?.args || {} },
1021
289
  auth: Function.isFunction(options?.auth) && options.auth,
1022
- chatType: new Set(options?.chatType || ['mention', PRIVATE]), // ignore GROUP, CHANNEL by default
290
+ bot: _bot,
291
+ chatType: new Set(options?.chatType || [MENTION, PRIVATE]), // ignore GROUP, CHANNEL by default
1023
292
  cmds: [...options?.cmds || []],
293
+ embed: options?.embed,
1024
294
  hello: options?.hello || HELLO,
1025
295
  help: { ...options?.help || {} },
1026
296
  homeGroup: options?.homeGroup,
1027
- info: options?.info || bot.lines([`[${EMOJI_BOT} ${pkg.title}](${pkg.homepage})`, pkg.description]),
1028
- magicWord: options?.magicWord && new Set(options.magicWord),
297
+ info: options?.info || bot.lines([`[${bot.BOT} ${pkg.title}](${pkg.homepage})`, pkg.description]),
298
+ lang: options?.lang,
299
+ pipeline: options?.pipeline || {},
1029
300
  private: options?.private && new Set(options.private),
1030
- session: { get: options?.session?.get, set: options?.session?.set },
1031
- skills: { ...options?.skills || {} },
1032
- speech: options?.speech,
1033
- vision: options?.vision,
1034
- supportedMimeTypes: options?.supportedMimeTypes || [],
1035
- database: options?.database,
1036
- memorize: options?.memorize,
1037
- embedding: options?.embedding,
301
+ rerank: options?.rerank,
302
+ storage: options?.storage,
303
+ supportedMimeTypes,
1038
304
  };
1039
- (!options?.session?.get || !options?.session?.set)
1040
- && log(`WARNING: Sessions persistence is not enabled.`);
1041
- const mods = [
1042
- ...subconscious.map(s => ({ ...s, run: s.run && !options?.silent })),
1043
- ...utilitas.ensureArray(options?.skill),
1044
- ];
1045
- if (hal._.database) {
1046
- assert(
1047
- [dbio.MYSQL, dbio.POSTGRESQL].includes(hal._.database?.provider),
1048
- 'Invalid database provider.'
1049
- );
1050
- assert(
1051
- hal._.database?.client?.query
1052
- && hal._.database?.client?.upsert,
1053
- 'Database client is required.'
1054
- );
1055
- const dbResult = [];
1056
- try {
1057
- for (const act of initSql[hal._.database?.provider]) {
1058
- dbResult.push(await hal._.database.client.query(...act));
1059
- }
1060
- } catch (e) { console.error(e); }
305
+ const mods = [];
306
+ if (_.storage) {
307
+ assert([
308
+ dbio.MYSQL, dbio.POSTGRESQL, storage.FILE
309
+ ].includes(_.storage?.provider), 'Invalid storage provider.');
310
+ if ([
311
+ dbio.MYSQL, dbio.POSTGRESQL
312
+ ].includes(_.storage?.provider)) {
313
+ assert(
314
+ _.storage?.client?.query && _.storage?.client?.upsert,
315
+ 'Invalid storage client.'
316
+ );
317
+ const dbResult = [];
318
+ try {
319
+ for (const act of initSql[_.storage?.provider]) {
320
+ dbResult.push(await _.storage.client.query(...act));
321
+ }
322
+ } catch (e) { console.error(e); }
323
+ }
1061
324
  }
1062
- for (let skillPath of utilitas.ensureArray(options?.skillPath)) {
1063
- log(`SKILLS: ${skillPath}`);
1064
- const files = (readdirSync(skillPath) || []).filter(
325
+ for (let pipeline of utilitas.ensureArray(options?.pipelinePath || [])) {
326
+ log(`PIPELINE: ${pipeline}`);
327
+ const files = (readdirSync(pipeline) || []).filter(
1065
328
  file => /\.mjs$/i.test(file) && !file.startsWith('.')
1066
329
  );
1067
330
  for (let f of files) {
1068
- const m = await import(join(skillPath, f));
331
+ const m = await import(join(pipeline, f));
1069
332
  mods.push({ ...m, name: m.name || f.replace(/^(.*)\.mjs$/i, '$1') });
1070
333
  }
1071
334
  }
1072
335
  mods.sort((x, y) => ~~x.priority - ~~y.priority).map(establish);
1073
- assert(mods.length, 'Invalid skill set.', 501);
336
+ assert(mods.length, 'Invalid pipeline.', 501);
1074
337
  await parseArgs(); // Validate args options.
1075
338
  }
1076
339
  }
1077
- assert(hal, 'Hal have not been initialized.', 501);
1078
- return hal;
340
+ assert(_, 'Bot have not been initialized.', 501);
341
+ return _;
1079
342
  };
1080
343
 
1081
-
1082
-
1083
344
  export default init;
1084
345
  export {
1085
- _NEED,
1086
346
  BINARY_STRINGS,
347
+ BOT_COMMAND,
348
+ CHANNEL,
349
+ CHECK,
1087
350
  COMMAND_DESCRIPTION_LENGTH,
1088
351
  COMMAND_LENGTH,
1089
352
  COMMAND_LIMIT,
1090
- EMOJI_BOT,
1091
- EMOJI_SPEECH,
1092
- GROUP_LIMIT,
353
+ GROUP,
1093
354
  HELLO,
1094
- PRIVATE_LIMIT,
355
+ MENTION,
356
+ OFF,
357
+ ON,
358
+ PRIVATE,
359
+ SEARCH_LIMIT,
360
+ parseArgs,
361
+ _,
1095
362
  end,
363
+ getContext,
1096
364
  init,
1097
- lines2,
365
+ json,
366
+ logOptions,
1098
367
  newCommand,
1099
368
  oList,
369
+ recall,
370
+ rerank,
371
+ table,
1100
372
  uList
1101
373
  };