Files
reverse-engineering-anything/src/domain/utf8Excerpt.ts
T

41 lines
1.3 KiB
TypeScript

import { excerptText, type ExcerptText } from "./excerpt.js";
/**
* The last `maximumBytes` UTF-8 bytes of `text`, never splitting a character.
* `bytes` is the excerpt's UTF-8 length; `truncated` says whether any of the
* start of `text` was dropped.
*/
export const utf8TailExcerpt = (
text: string,
maximumBytes: number,
): {
readonly text: ExcerptText;
readonly bytes: number;
readonly truncated: boolean;
} => {
// A UTF-16 suffix needs at least one UTF-8 byte per code unit.
// Include one extra unit so the candidate cannot cut a surrogate pair
// that would fit in the result.
const candidateStart = Math.max(0, text.length - Math.ceil(maximumBytes) - 1);
const encoded = Buffer.from(text.slice(candidateStart), "utf8");
if (candidateStart === 0 && encoded.byteLength <= maximumBytes)
return {
text: excerptText(text),
bytes: encoded.byteLength,
truncated: false,
};
let start = Math.max(
0,
encoded.byteLength - Math.max(0, Math.floor(maximumBytes)),
);
// UTF-8 continuation bytes are 0b10xxxxxx; start at the next character.
while (start < encoded.byteLength && ((encoded[start] ?? 0) & 0xc0) === 0x80)
start += 1;
const tail = encoded.subarray(start);
return {
text: excerptText(tail.toString("utf8")),
bytes: tail.byteLength,
truncated: true,
};
};