99 lines
2.6 KiB
TypeScript
99 lines
2.6 KiB
TypeScript
const compactWhitespace = (value: string) => value.replace(/\s+/g, " ").trim();
|
||
const MAX_SPEECH_CHUNK_LENGTH = 520;
|
||
const MIN_SPEECH_CHUNK_LENGTH = 180;
|
||
const SPEECH_SENTENCE_PATTERN = /[^。！？!?；;\n]+(?:[。！？!?；;]+|(?=\n|$))/g;
|
||
|
||
const normalizeWithOffsetMap = (value: string) => {
|
||
let normalized = "";
|
||
const offsetMap: number[] = [];
|
||
let isPreviousWhitespace = false;
|
||
|
||
Array.from(value).forEach((char, index) => {
|
||
if (/\s/u.test(char)) {
|
||
if (!isPreviousWhitespace && normalized.length > 0) {
|
||
normalized += " ";
|
||
offsetMap.push(index);
|
||
}
|
||
isPreviousWhitespace = true;
|
||
return;
|
||
}
|
||
|
||
normalized += char;
|
||
offsetMap.push(index);
|
||
isPreviousWhitespace = false;
|
||
});
|
||
|
||
return {
|
||
normalized: normalized.trimEnd(),
|
||
offsetMap,
|
||
};
|
||
};
|
||
|
||
export function findSpeechSelectionStartOffset(
|
||
text: string,
|
||
selectedText: string,
|
||
): number | null {
|
||
const needle = selectedText.trim();
|
||
if (!needle) return null;
|
||
|
||
const exactIndex = text.indexOf(needle);
|
||
if (exactIndex >= 0) return exactIndex;
|
||
|
||
const normalizedNeedle = compactWhitespace(needle);
|
||
if (!normalizedNeedle) return null;
|
||
|
||
const haystack = normalizeWithOffsetMap(text);
|
||
const normalizedIndex = haystack.normalized.indexOf(normalizedNeedle);
|
||
if (normalizedIndex < 0) return null;
|
||
|
||
return haystack.offsetMap[normalizedIndex] ?? null;
|
||
}
|
||
|
||
export function splitSpeechTextIntoChunks(text: string): string[] {
|
||
const normalizedText = text.trim();
|
||
if (!normalizedText) return [];
|
||
|
||
const segments = Array.from(normalizedText.matchAll(SPEECH_SENTENCE_PATTERN), (match) =>
|
||
compactWhitespace(match[0]),
|
||
).filter(Boolean);
|
||
const sourceSegments = segments.length > 0 ? segments : [normalizedText];
|
||
const chunks: string[] = [];
|
||
let currentChunk = "";
|
||
|
||
const flush = () => {
|
||
if (!currentChunk) return;
|
||
chunks.push(currentChunk);
|
||
currentChunk = "";
|
||
};
|
||
|
||
const pushLongSegment = (segment: string) => {
|
||
for (let offset = 0; offset < segment.length; offset += MAX_SPEECH_CHUNK_LENGTH) {
|
||
chunks.push(segment.slice(offset, offset + MAX_SPEECH_CHUNK_LENGTH));
|
||
}
|
||
};
|
||
|
||
sourceSegments.forEach((segment) => {
|
||
if (segment.length > MAX_SPEECH_CHUNK_LENGTH) {
|
||
flush();
|
||
pushLongSegment(segment);
|
||
return;
|
||
}
|
||
|
||
const candidate = currentChunk ? `${currentChunk}${segment}` : segment;
|
||
if (
|
||
currentChunk &&
|
||
candidate.length > MAX_SPEECH_CHUNK_LENGTH &&
|
||
currentChunk.length >= MIN_SPEECH_CHUNK_LENGTH
|
||
) {
|
||
flush();
|
||
currentChunk = segment;
|
||
return;
|
||
}
|
||
|
||
currentChunk = candidate;
|
||
});
|
||
|
||
flush();
|
||
return chunks;
|
||
}
|