a collection of lightweight TypeScript packages for AT Protocol, the protocol powering Bluesky
atproto
bluesky
typescript
npm
fix(lexicon-doc): faulty UTF-16 fast-path constraint validation
+14
-7
+5
.changeset/yellow-cycles-warn.md
+5
.changeset/yellow-cycles-warn.md
+9
-7
packages/lexicons/lexicon-doc/lib/utils.ts
+9
-7
packages/lexicons/lexicon-doc/lib/utils.ts
···
1
1
const segmenter = new Intl.Segmenter();
2
2
3
3
export const isWithinUtf8Bounds = (input: string, min = 0, max = Infinity): 'max' | 'min' | undefined => {
4
-
const maybeUtf8Len = input.length * 3;
4
+
const utf16Len = input.length;
5
+
const maybeUtf8Len = utf16Len * 3;
5
6
6
-
// fail early if we're still less than minimum length
7
+
// fail early if estimated upper bound is too small
7
8
if (maybeUtf8Len < min) {
8
9
return 'min';
9
10
}
10
11
11
-
// skip if we're still within maximum length
12
-
if (maybeUtf8Len <= max) {
12
+
// skip if UTF-16 length already satisfies both constraints
13
+
if (utf16Len >= min && maybeUtf8Len <= max) {
13
14
return undefined;
14
15
}
15
16
···
32
33
33
34
const utf16Len = input.length;
34
35
35
-
// fail early if UTF-16 length is less than grapheme length
36
+
// fail early if UTF-16 length is too small
36
37
if (utf16Len < min) {
37
38
return 'min';
38
39
}
39
40
40
-
// skip if we're still within maximum constraint
41
-
if (utf16Len <= max) {
41
+
// if there is no minimum bounds, we can safely skip when UTF-16 is
42
+
// within the maximum bounds.
43
+
if (min === 0 && utf16Len <= max) {
42
44
return undefined;
43
45
}
44
46