To Consider

[GeorgeKutivadze]

🚨 TL;DR

Your fetch fails because:

urlEncode is broken for non-ASCII (Japanese) characters.

artist_name contains noisy metadata like CV: and feat. which hurts matching.

You are not using ISRC (which you actually have).

Fix:

Replace urlEncode with net/url.QueryEscape

Normalize artist name before request

Prefer ISRC search if available

🔴 Root Cause #1 — Your urlEncode() is wrong

This is the real bug.

default:
b.WriteString(fmt.Sprintf("%%%02X", r))

r is a Unicode rune, not a byte.

For:

#超絶かわいい

You are generating invalid percent encoding.

LRCLIB expects UTF-8 percent encoded bytes, not rune hex.

✅ FIX 1 — Replace urlEncode

Delete your entire function and use standard lib.

import "net/url"

Replace URI building with:

params := url.Values{}
params.Set("track_name", track.Title)
params.Set("artist_name", normalizeArtist(track.Artist))
params.Set("album_name", track.Album)
params.Set("duration", strconv.Itoa(duration))

uri := fmt.Sprintf("%s/api/get?%s", baseURL, params.Encode())

Delete urlEncode() completely.

🔴 Root Cause #2 — Artist Name Is Dirty

You are sending:

mona(CV:夏川椎菜) feat. HoneyWorks

LRCLIB likely stores:

mona
or
mona feat. HoneyWorks

The CV: part is poisoning the match.

✅ FIX 2 — Normalize Artist

Add this:

func normalizeArtist(s string) string {
s = strings.ReplaceAll(s, "CV:", "")
s = strings.ReplaceAll(s, "CV", "")
s = strings.ReplaceAll(s, "（", "(")
s = strings.ReplaceAll(s, "）", ")")

if i := strings.Index(s, "("); i != -1 {
    s = s[:i]
}

if i := strings.Index(strings.ToLower(s), " feat."); i != -1 {
    s = s[:i]
}

return strings.TrimSpace(s)

}

Then:

params.Set("artist_name", normalizeArtist(track.Artist))
🟢 Root Cause #3 — You Have ISRC (Use It!)

You have:

isrc: JPX402200177

LRCLIB supports lookup by ISRC.

This is far more reliable for Japanese tracks.

✅ FIX 3 — Prefer ISRC Lookup

Modify fetchLyrics:

GET /api/get?isrc=JPX402200177

You need to:

Extend jobInfo to store ISRC

Fetch it from Subsonic song tags

If ISRC exists → use it instead of title/artist

Example:

if track.ISRC != "" {
uri = fmt.Sprintf("%s/api/get?isrc=%s",
baseURL,
url.QueryEscape(track.ISRC),
)
}
🔥 Why Japanese Was Failing Specifically

Because:

Japanese = multi-byte UTF-8

Your encoder assumes 1 rune = 1 byte

That produces invalid percent sequences

LRCLIB rejects it → 404

ASCII tracks would work.
Japanese tracks fail.

✅ Minimal Required Fix

If you only do ONE thing:

Replace urlEncode() with url.QueryEscape.

That alone will fix most failures.

🧪 After Fix — Your Example Will Work

For:

Track: #超絶かわいい
Artist: mona(CV:夏川椎菜) feat. HoneyWorks

Proper request becomes:

...track_name=%23%E8%B6%85%E7%B5%B6%E3%81%8B%E3%82%8F%E3%81%84%E3%81%84

Correct UTF-8 percent encoding.

🚀 Recommended Final Improvements

Use url.Values

Normalize artist

Prefer ISRC lookup

Log final query string for debugging