fork(fix): Clone AMLL 并修复 BUG

- 将AMLL Clone到本以地进行修复和优化(emm虽然这很不优雅但是暂时无时间做子模块和Fork)
- 修复在当前播放歌词行不可见的视口Seek会出现滚动偏移的问题

amll-local/packages/lyric/src/formats/yrc.ts

@@ -0,0 +1,140 @@
+/**
+ * @fileoverview YRC（网易云音乐逐词歌词）格式解析与生成。
+ * 行开头为 [startTime,duration]，每个词为 (startTime,duration,0)word
+ *
+ * 格式示例：
+ * [190871,1984](190871,361,0)For (191232,172,0)the (191404,376,0)first (191780,1075,0)time
+ * [193459,4198](193459,412,0)What's (193871,574,0)past (194445,506,0)is (194951,2706,0)past
+ */
+import type { LyricLine, LyricWord } from "../types";
+import {
+	createLine,
+	createWord,
+	normalizeDuration,
+	normalizeTimestamp,
+} from "../utils";
+
+const beginParenPattern = /^[（(]/;
+const endParenPattern = /[）)]$/;
+function checkIsBG(words: LyricWord[]): boolean {
+	return (
+		words.length > 0 &&
+		beginParenPattern.test(words[0].word) &&
+		endParenPattern.test(words[words.length - 1].word)
+	);
+}
+function trimBGParentheses(words: LyricWord[]): void {
+	words[0].word = words[0].word.slice(1);
+	words[words.length - 1].word = words[words.length - 1].word.slice(0, -1);
+}
+
+/**
+ * 解析 YRC 格式的歌词字符串
+ * @param yrc 歌词字符串
+ * @returns 成功解析出来的歌词
+ */
+export function parseYrc(yrc: string): LyricLine[] {
+	const wordPattern = /^(.*?)\((\d+),(\d+),0\)/;
+	const linePattern = /^\[(\d+),(\d+)\]/;
+
+	const lines = yrc
+		.split(/\r?\n/)
+		.map((l) => l.trim())
+		.filter((l) => l.length > 0);
+
+	return lines
+		.map((lineStr) => {
+			const lineMatch = lineStr.match(linePattern);
+			if (!lineMatch) return null;
+			const [linePrefix, lineStartStr, lineDurStr] = lineMatch;
+			const lineStart = Number(lineStartStr);
+			const lineDuration = Number(lineDurStr);
+
+			const words: LyricWord[] = [];
+			let lineContent = lineStr.slice(linePrefix.length).trim();
+			if (!lineContent) return null;
+
+			let lastStart = -1;
+			let lastEnd = -1;
+			while (true) {
+				const wordMatch = lineContent.match(wordPattern);
+				if (!wordMatch) break;
+				const [fullMatch, lastText, wordStartStr, wordDurStr] = wordMatch;
+				if (lastText && lastStart !== -1)
+					words.push(
+						createWord({
+							word: lastText,
+							startTime: lastStart,
+							endTime: lastEnd,
+						}),
+					);
+				const wordStart = Number(wordStartStr);
+				const wordDur = Number(wordDurStr);
+				const wordEnd = wordStart + wordDur;
+				[lastStart, lastEnd] = [wordStart, wordEnd];
+				lineContent = lineContent.slice(fullMatch.length);
+			}
+			if (lastStart !== -1 && lineContent)
+				words.push(
+					createWord({
+						word: lineContent,
+						startTime: lastStart,
+						endTime: lastEnd,
+					}),
+				);
+
+			const isBG = checkIsBG(words);
+			if (isBG) trimBGParentheses(words);
+			return createLine({
+				startTime: lineStart,
+				endTime: lineStart + lineDuration,
+				words,
+				isBG,
+			});
+		})
+		.filter((line): line is LyricLine => line !== null);
+}
+
+function makeParenthesesFull(text: string): string {
+	return text.replace(/\(/g, "（").replace(/\)/g, "）");
+}
+
+/**
+ * 将歌词数组转换为 YRC 格式的字符串
+ * @param lines 歌词数组
+ * @returns YRC 格式的字符串
+ */
+export function stringifyYrc(lines: LyricLine[]): string {
+	return lines
+		.map((line) => {
+			const lineStart = normalizeTimestamp(line.startTime);
+			const lineEnd = normalizeTimestamp(line.endTime);
+			const lineDuration = normalizeDuration(lineEnd - lineStart);
+
+			const lineWords: string[] = [];
+			for (const [
+				index,
+				{ word, startTime, endTime },
+			] of line.words.entries()) {
+				if (!word.trim() && lineWords.length) {
+					lineWords[lineWords.length - 1] += word;
+					continue;
+				}
+				let printedWord = makeParenthesesFull(word);
+				if (line.isBG) {
+					if (index === 0) printedWord = `（${printedWord}`;
+					if (index === line.words.length - 1) printedWord += "）";
+				}
+				const normalizedWordStart = normalizeTimestamp(startTime);
+				const normalizedWordEnd = normalizeTimestamp(endTime);
+				const wordDuration = normalizeDuration(
+					normalizedWordEnd - normalizedWordStart,
+				);
+				lineWords.push(
+					`(${normalizedWordStart},${wordDuration},0)${printedWord}`,
+				);
+			}
+			return `[${lineStart},${lineDuration}]${lineWords.join("")}`;
+		})
+		.join("\n");
+}