perf(lyrics): drop []byte/string round-trips in parsers

Apply code-review feedback to remove avoidable allocations in the lyrics
parsers. isTTMLDocument now takes []byte directly, so parseTTMLWithDefaultLang
no longer copies its buffer into a string before the TTML probe. parseSRTBlock
splits its block with strings.Split instead of converting to []byte and back
per line. ParseLyrics hoists strings.ToLower(suffix) out of the format loop.

No behavior change; the dropped len(scanner)==0 SRT guard was dead (strings.Split
never returns an empty slice, and the existing len(lines)==0 check still covers
empty input).

Signed-off-by: Deluan <deluan@navidrome.org>

model/lyrics_parse.go

@@ -30,13 +30,14 @@ var lyricFormats = []struct {
 // a structured parser that does not match falls back to the LRC/plain-text floor.
 func ParseLyrics(suffix, lang string, contents []byte) (LyricList, error) {
 	contents = stripBOM(contents)
-	sniff := suffix == "" || strings.EqualFold(suffix, "auto")
+	suffix = strings.ToLower(suffix)
+	sniff := suffix == "" || suffix == "auto"
 
 	// Sniffing tries every format in order; a known suffix selects just its own.
 	// Unmatched suffixes leave no candidates, so parseFirstMatch falls to plain.
 	candidates := make([]lyricParser, 0, len(lyricFormats))
 	for _, f := range lyricFormats {
-		if sniff || slices.Contains(f.suffixes, strings.ToLower(suffix)) {
+		if sniff || slices.Contains(f.suffixes, suffix) {
 			candidates = append(candidates, f.parse)
 		}
 	}

model/lyrics_srt.go

@@ -1,7 +1,6 @@
 package model
 
 import (
-	"bytes"
 	"regexp"
 	"strconv"
 	"strings"
@@ -61,14 +60,10 @@ func splitSRTBlocks(raw string) []string {
 }
 
 func parseSRTBlock(block string) (Line, bool, error) {
-	scanner := bytes.Split([]byte(block), []byte("\n"))
-	if len(scanner) == 0 {
-		return Line{}, false, nil
-	}
-
-	lines := make([]string, 0, len(scanner))
-	for _, line := range scanner {
-		lines = append(lines, strings.TrimSpace(string(line)))
+	rawLines := strings.Split(block, "\n")
+	lines := make([]string, 0, len(rawLines))
+	for _, line := range rawLines {
+		lines = append(lines, strings.TrimSpace(line))
 	}
 
 	if len(lines) == 0 {

model/lyrics_ttml.go

@@ -102,8 +102,8 @@ type ttmlParser struct {
 	metadataSeq int
 }
 
-func isTTMLDocument(text string) bool {
-	decoder := xml.NewDecoder(strings.NewReader(strings.TrimSpace(text)))
+func isTTMLDocument(contents []byte) bool {
+	decoder := xml.NewDecoder(bytes.NewReader(bytes.TrimSpace(contents)))
 	for {
 		token, err := decoder.Token()
 		if err != nil {
@@ -120,7 +120,7 @@ func parseTTMLWithDefaultLang(defaultLang string, contents []byte) (LyricList, e
 
 	// Skip non-TTML content so sniffing doesn't run the XML decoder on plain text.
 	// Checked after the encoding fixup so UTF-16-declared documents are recognized.
-	if !isTTMLDocument(string(contents)) {
+	if !isTTMLDocument(contents) {
 		return nil, nil
 	}