a collection of lightweight TypeScript packages for AT Protocol, the protocol powering Bluesky
atproto
bluesky
typescript
npm
feat(bluesky-richtext-parser): parse cashtag syntax
+222
-1
+5
.changeset/breezy-geese-drum.md
+5
.changeset/breezy-geese-drum.md
+193
packages/bluesky/richtext-parser/lib/index.test.ts
+193
packages/bluesky/richtext-parser/lib/index.test.ts
···
437
437
]);
438
438
});
439
439
440
+
it('cashtags', () => {
441
+
expect(tokenize('$AAPL')).toEqual([
442
+
{
443
+
name: 'AAPL',
444
+
raw: '$AAPL',
445
+
type: 'cashtag',
446
+
},
447
+
]);
448
+
449
+
expect(tokenize('＄AAPL')).toEqual([
450
+
{
451
+
name: 'AAPL',
452
+
raw: '＄AAPL',
453
+
type: 'cashtag',
454
+
},
455
+
]);
456
+
457
+
expect(tokenize('$aapl')).toEqual([
458
+
{
459
+
name: 'aapl',
460
+
raw: '$aapl',
461
+
type: 'cashtag',
462
+
},
463
+
]);
464
+
465
+
expect(tokenize('$BTC')).toEqual([
466
+
{
467
+
name: 'BTC',
468
+
raw: '$BTC',
469
+
type: 'cashtag',
470
+
},
471
+
]);
472
+
473
+
expect(tokenize('$DOGE2')).toEqual([
474
+
{
475
+
name: 'DOGE2',
476
+
raw: '$DOGE2',
477
+
type: 'cashtag',
478
+
},
479
+
]);
480
+
481
+
expect(tokenize('$LONGNAME')).toEqual([
482
+
{
483
+
name: 'LONGNAME',
484
+
raw: '$LONGNAME',
485
+
type: 'cashtag',
486
+
},
487
+
]);
488
+
489
+
// too long - doesn't match (no boundary after 8 chars)
490
+
expect(tokenize('$TOOLONGNAME')).toEqual([
491
+
{
492
+
content: '$TOOLONGNAME',
493
+
raw: '$TOOLONGNAME',
494
+
type: 'text',
495
+
},
496
+
]);
497
+
498
+
// must start with letter
499
+
expect(tokenize('$123')).toEqual([
500
+
{
501
+
content: '$123',
502
+
raw: '$123',
503
+
type: 'text',
504
+
},
505
+
]);
506
+
507
+
expect(tokenize('hello$AAPL')).toEqual([
508
+
{
509
+
content: 'hello$AAPL',
510
+
raw: 'hello$AAPL',
511
+
type: 'text',
512
+
},
513
+
]);
514
+
515
+
expect(tokenize('hello($AAPL')).toEqual([
516
+
{
517
+
content: 'hello(',
518
+
raw: 'hello(',
519
+
type: 'text',
520
+
},
521
+
{
522
+
name: 'AAPL',
523
+
raw: '$AAPL',
524
+
type: 'cashtag',
525
+
},
526
+
]);
527
+
528
+
expect(tokenize('$AAPL.')).toEqual([
529
+
{
530
+
name: 'AAPL',
531
+
raw: '$AAPL',
532
+
type: 'cashtag',
533
+
},
534
+
{
535
+
content: '.',
536
+
raw: '.',
537
+
type: 'text',
538
+
},
539
+
]);
540
+
541
+
expect(tokenize('$AAPL$')).toEqual([
542
+
{
543
+
content: '$AAPL$',
544
+
raw: '$AAPL$',
545
+
type: 'text',
546
+
},
547
+
]);
548
+
549
+
expect(tokenize('hello $AAPL$')).toEqual([
550
+
{
551
+
content: 'hello ',
552
+
raw: 'hello $AAPL$',
553
+
type: 'text',
554
+
},
555
+
]);
556
+
557
+
expect(tokenize('$AAPL.$')).toEqual([
558
+
{
559
+
name: 'AAPL',
560
+
raw: '$AAPL',
561
+
type: 'cashtag',
562
+
},
563
+
{
564
+
content: '.',
565
+
raw: '.$',
566
+
type: 'text',
567
+
},
568
+
]);
569
+
570
+
expect(tokenize('$AAPL hello')).toEqual([
571
+
{
572
+
name: 'AAPL',
573
+
raw: '$AAPL',
574
+
type: 'cashtag',
575
+
},
576
+
{
577
+
content: ' hello',
578
+
raw: ' hello',
579
+
type: 'text',
580
+
},
581
+
]);
582
+
583
+
expect(tokenize('hello $AAPL hello')).toEqual([
584
+
{
585
+
content: 'hello ',
586
+
raw: 'hello ',
587
+
type: 'text',
588
+
},
589
+
{
590
+
name: 'AAPL',
591
+
raw: '$AAPL',
592
+
type: 'cashtag',
593
+
},
594
+
{
595
+
content: ' hello',
596
+
raw: ' hello',
597
+
type: 'text',
598
+
},
599
+
]);
600
+
601
+
expect(tokenize('hello $AAPL')).toEqual([
602
+
{
603
+
content: 'hello ',
604
+
raw: 'hello ',
605
+
type: 'text',
606
+
},
607
+
{
608
+
name: 'AAPL',
609
+
raw: '$AAPL',
610
+
type: 'cashtag',
611
+
},
612
+
]);
613
+
614
+
expect(tokenize('$AAPL $MSFT')).toEqual([
615
+
{
616
+
name: 'AAPL',
617
+
raw: '$AAPL',
618
+
type: 'cashtag',
619
+
},
620
+
{
621
+
content: ' ',
622
+
raw: ' ',
623
+
type: 'text',
624
+
},
625
+
{
626
+
name: 'MSFT',
627
+
raw: '$MSFT',
628
+
type: 'cashtag',
629
+
},
630
+
]);
631
+
});
632
+
440
633
it('autolinks', () => {
441
634
expect(tokenize('https://example.com')).toEqual([
442
635
{
+24
-1
packages/bluesky/richtext-parser/lib/index.ts
+24
-1
packages/bluesky/richtext-parser/lib/index.ts
···
5
5
const TOPIC_RE =
6
6
/^(?:#(?!\ufe0f|\u20e3)|＃)([\p{N}]*[\p{L}\p{M}\p{Pc}][\p{L}\p{M}\p{Pc}\p{N}]*)($|\s|\p{P})/u;
7
7
8
+
const CASHTAG_RE = /^[$＄]([A-Za-z][A-Za-z0-9]{0,7})($|\s|\p{P})/u;
9
+
8
10
const EMOTE_RE = /^:([\w-]+):/;
9
11
10
12
const AUTOLINK_RE = /^https?:\/\/[\S]+/;
···
26
28
const CODE_RE = /^(`+)([^]*?[^`])\1(?!`)/;
27
29
const CODE_ESCAPE_BACKTICKS_RE = /^ (?= *`)|(` *) $/g;
28
30
29
-
const TEXT_RE = /^[^]+?(?:(?=$|[~*_`:\\[]|https?:\/\/)|(?<=\s|[(){}/\\[\]\-|:;'".,=+])(?=[@＠#＃]))/;
31
+
const TEXT_RE = /^[^]+?(?:(?=$|[~*_`:\\[]|https?:\/\/)|(?<=\s|[(){}/\\[\]\-|:;'".,=+])(?=[@＠#＃$＄]))/;
30
32
31
33
export interface EscapeToken {
32
34
type: 'escape';
···
46
48
name: string;
47
49
}
48
50
51
+
export interface CashtagToken {
52
+
type: 'cashtag';
53
+
raw: string;
54
+
name: string;
55
+
}
56
+
49
57
export interface EmoteToken {
50
58
type: 'emote';
51
59
raw: string;
···
105
113
| EscapeToken
106
114
| MentionToken
107
115
| TopicToken
116
+
| CashtagToken
108
117
| EmoteToken
109
118
| AutolinkToken
110
119
| LinkToken
···
152
161
}
153
162
};
154
163
164
+
const tokenizeCashtag = (src: string): CashtagToken | undefined => {
165
+
const match = CASHTAG_RE.exec(src);
166
+
if (match && match[2] !== '$') {
167
+
const suffix = match[2].length;
168
+
169
+
return {
170
+
type: 'cashtag',
171
+
raw: suffix > 0 ? match[0].slice(0, -suffix) : match[0],
172
+
name: match[1],
173
+
};
174
+
}
175
+
};
176
+
155
177
const tokenizeEmote = (src: string): EmoteToken | undefined => {
156
178
const match = EMOTE_RE.exec(src);
157
179
if (match) {
···
296
318
tokenizeAutolink(src) ||
297
319
tokenizeMention(src) ||
298
320
tokenizeTopic(src) ||
321
+
tokenizeCashtag(src) ||
299
322
tokenizeEmote(src) ||
300
323
tokenizeLink(src) ||
301
324
tokenizeEmStrongU(src) ||