a collection of lightweight TypeScript packages for AT Protocol, the protocol powering Bluesky
atproto
bluesky
typescript
npm
refactor(util-text): add ASCII fast-path to grapheme counting
+79
+5
.changeset/ten-islands-attack.md
+5
.changeset/ten-islands-attack.md
+37
packages/misc/util-text/lib/index.rn.ts
+37
packages/misc/util-text/lib/index.rn.ts
···
1
1
import { countGraphemes } from 'unicode-segmenter/grapheme';
2
2
3
+
const isAsciiWithoutCr = (text: string): boolean => {
4
+
const len = text.length;
5
+
let idx = 0;
6
+
7
+
while (idx + 3 < len) {
8
+
const a = text.charCodeAt(idx);
9
+
const b = text.charCodeAt(idx + 1);
10
+
const c = text.charCodeAt(idx + 2);
11
+
const d = text.charCodeAt(idx + 3);
12
+
13
+
if ((a | b | c | d) > 0x7f || a === 0x0d || b === 0x0d || c === 0x0d || d === 0x0d) {
14
+
return false;
15
+
}
16
+
17
+
idx += 4;
18
+
}
19
+
20
+
while (idx < len) {
21
+
const code = text.charCodeAt(idx);
22
+
if (code > 0x7f || code === 0x0d) {
23
+
return false;
24
+
}
25
+
26
+
idx++;
27
+
}
28
+
29
+
return true;
30
+
};
31
+
3
32
/**
4
33
* returns the grapheme length of a string
5
34
* @param text string to count graphemes in
6
35
* @returns grapheme count
7
36
*/
8
37
export const getGraphemeLength = (text: string): number => {
38
+
if (isAsciiWithoutCr(text)) {
39
+
return text.length;
40
+
}
41
+
9
42
return countGraphemes(text);
10
43
};
11
44
···
28
61
// grapheme count is definitely within max
29
62
if (min === 0 && utf16Len <= max) {
30
63
return true;
64
+
}
65
+
66
+
if (isAsciiWithoutCr(text)) {
67
+
return utf16Len <= max;
31
68
}
32
69
33
70
const count = countGraphemes(text);
+37
packages/misc/util-text/lib/index.ts
+37
packages/misc/util-text/lib/index.ts
···
1
1
const segmenter = new Intl.Segmenter();
2
2
3
+
const isAsciiWithoutCr = (text: string): boolean => {
4
+
const len = text.length;
5
+
let idx = 0;
6
+
7
+
while (idx + 3 < len) {
8
+
const a = text.charCodeAt(idx);
9
+
const b = text.charCodeAt(idx + 1);
10
+
const c = text.charCodeAt(idx + 2);
11
+
const d = text.charCodeAt(idx + 3);
12
+
13
+
if ((a | b | c | d) > 0x7f || a === 0x0d || b === 0x0d || c === 0x0d || d === 0x0d) {
14
+
return false;
15
+
}
16
+
17
+
idx += 4;
18
+
}
19
+
20
+
while (idx < len) {
21
+
const code = text.charCodeAt(idx);
22
+
if (code > 0x7f || code === 0x0d) {
23
+
return false;
24
+
}
25
+
26
+
idx++;
27
+
}
28
+
29
+
return true;
30
+
};
31
+
3
32
/**
4
33
* returns the grapheme length of a string
5
34
* @param text string to count graphemes in
6
35
* @returns grapheme count
7
36
*/
8
37
export const getGraphemeLength = (text: string): number => {
38
+
if (isAsciiWithoutCr(text)) {
39
+
return text.length;
40
+
}
41
+
9
42
const iterator = segmenter.segment(text)[Symbol.iterator]();
10
43
let count = 0;
11
44
···
35
68
// grapheme count is definitely within max
36
69
if (min === 0 && utf16Len <= max) {
37
70
return true;
71
+
}
72
+
73
+
if (isAsciiWithoutCr(text)) {
74
+
return utf16Len <= max;
38
75
}
39
76
40
77
// count graphemes with early termination