LrcParser.kt

52 lines · 2.2 KB

Changed in: 0.5.0

1	package dev.stave.lyrics
2
3	/**
4	* Parser for the LRC subtitle format as used by LRCLIB.
5	*
6	* Handles standard `[mm:ss.xx]text` lines, multi-timestamp lines
7	* (`[00:12.34][00:45.67]chorus`) which expand into multiple LrcLines with the same
8	* text, metadata tags (`[ar:…]`, `[ti:…]`) which are discarded, and malformed lines
9	* which are silently skipped.
10	*
11	* Returns lines sorted ascending by timestamp.
12	*/
13	object LrcParser {
14
15	// `[mm:ss.xx]` or `[mm:ss.xxx]`. mm can be 2+ digits for tracks over 99 minutes.
16	private val TIMESTAMP = Regex("""\[(\d{1,3}):(\d{2})(?:\.(\d{1,3}))?\]""")
17
18	// `[ar:…]`, `[ti:…]`, `[length:…]` — metadata tags we discard.
19	private val META_TAG = Regex("""\[[a-zA-Z]+:[^\]]*\]""")
20
21	fun parse(lrc: String): List<LrcLine> {
22	val out = mutableListOf<LrcLine>()
23	for (raw in lrc.lineSequence()) {
24	val line = raw.trim()
25	if (line.isBlank() \|\| META_TAG.matchEntire(line) != null) continue
26
27	val stamps = TIMESTAMP.findAll(line).toList()
28	if (stamps.isEmpty()) continue
29
30	// Text is whatever comes after the last timestamp on the line.
31	val lastStamp = stamps.last()
32	val text = line.substring(lastStamp.range.last + 1).trim()
33	// Skip lines with empty text — some sources pad with empty timestamp
34	// markers to align timing with instrumental sections.
35	if (text.isBlank()) continue
36
37	for (m in stamps) {
38	val mm = m.groupValues[1].toIntOrNull() ?: continue
39	val ss = m.groupValues[2].toIntOrNull() ?: continue
40	val fracStr = m.groupValues[3]
41	// Normalize fractional to 3 digits. "34" → 340, "345" → 345, "" → 0.
42	val frac = if (fracStr.isBlank()) 0 else {
43	val padded = fracStr.padEnd(3, '0').take(3)
44	padded.toIntOrNull() ?: 0
45	}
46	val millis = (mm * 60_000L) + (ss * 1000L) + frac.toLong()
47	if (millis >= 0) out.add(LrcLine.at(millis, text))
48	}
49	}
50	return out.sortedBy { it.start }
51	}
52	}