fix(server/m7-382): strip NUL bytes from MBIDs before persist

Follow-up to 947f944. With the dhowden/tag/mbz extractor now correctly
matching ID3v2 TXXX frames, mbid backfill started rejecting every row
with "invalid byte sequence for encoding UTF8: 0x00 (SQLSTATE 22021)".

Cause: dhowden's readTextWithDescrFrame (id3v2frames.go:454) — the path
that decodes TXXX text — does NOT strip the trailing/embedded NUL bytes
that ID3v2.4 uses as frame terminator and multi-value separator. Unlike
readTFrame, which does (line 313). So a TXXX:MusicBrainz Album Id of
"abc-123\x00" comes back to us verbatim, and Postgres refuses to store
NULs in text columns.

cleanMBID splits on NUL and returns the first non-empty segment, which
also handles the multi-value case (collaboration artist IDs are
NUL-separated; Minstrel uses the primary for MBCAA, so first wins).

Tests cover the trailing-NUL, multi-value, and all-NUL paths.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

internal/library/mbids.go

@@ -1,6 +1,8 @@
 package library
 
 import (
+	"strings"
+
 	"github.com/dhowden/tag"
 	"github.com/dhowden/tag/mbz"
 )
@@ -13,11 +15,26 @@ import (
 //     a *tag.Comm value whose Description holds the Picard tag name.
 //   - MP4 (M4A) freeform iTunes atoms are stored under their bare sub-name.
 //
-// We previously hand-rolled this lookup against guessed keys
+// Then sanitizes: dhowden's readTextWithDescrFrame (used for ID3v2 TXXX)
-// ("MUSICBRAINZ_ALBUMID", "TXXX:MusicBrainz Album Id", ...), none of which
+// does NOT strip the trailing/embedded NUL bytes that ID3v2.4 uses as
-// dhowden actually produces — so every extraction returned ("",""), which
+// frame terminator and multi-value separator — so the Text often comes
-// fed an empty MBID into both the scanner and the boot-time backfill.
+// out as "uuid\x00" or "uuid1\x00uuid2". Postgres rejects NULs in text
+// columns (SQLSTATE 22021). We split on NUL and take the first non-empty
+// segment — Minstrel uses the primary release artist for MBCAA lookup,
+// so first-wins matches the original semantics.
 func extractMBIDs(m tag.Metadata) (albumMBID, artistMBID string) {
 	info := mbz.Extract(m)
-	return info.Get(mbz.Album), info.Get(mbz.Artist)
+	return cleanMBID(info.Get(mbz.Album)), cleanMBID(info.Get(mbz.Artist))
+}
+
+func cleanMBID(s string) string {
+	if s == "" {
+		return ""
+	}
+	for _, part := range strings.Split(s, "\x00") {
+		if t := strings.TrimSpace(part); t != "" {
+			return t
+		}
+	}
+	return ""
 }

internal/library/mbids_test.go

@@ -109,3 +109,57 @@ func TestExtractMBIDs_VorbisUppercaseKeysMiss(t *testing.T) {
 		t.Errorf("uppercase Vorbis keys must not match dhowden output; got (%q, %q)", a, ar)
 	}
 }
+
+func TestExtractMBIDs_ID3v2StripsTrailingNUL(t *testing.T) {
+	// dhowden's readTextWithDescrFrame leaves the ID3v2.4 frame-
+	// terminator NUL on TXXX values. Postgres rejects NULs in text
+	// columns (SQLSTATE 22021), so we must strip them.
+	m := stubMeta{
+		format: tag.ID3v2_4,
+		raw: map[string]interface{}{
+			"TXXX": &tag.Comm{
+				Description: "MusicBrainz Album Id",
+				Text:        "abc-123\x00",
+			},
+			"TXXX_0": &tag.Comm{
+				Description: "MusicBrainz Artist Id",
+				Text:        "def-456\x00",
+			},
+		},
+	}
+	a, ar := extractMBIDs(m)
+	if a != "abc-123" || ar != "def-456" {
+		t.Errorf("got (%q, %q), want trailing NUL stripped", a, ar)
+	}
+}
+
+func TestExtractMBIDs_ID3v2MultiValueTakesFirst(t *testing.T) {
+	// Some Picard configs emit multi-value TXXX:MusicBrainz Artist Id
+	// for collaborations, NUL-separated. Minstrel uses the primary
+	// release artist for MBCAA, so first wins.
+	m := stubMeta{
+		format: tag.ID3v2_4,
+		raw: map[string]interface{}{
+			"TXXX": &tag.Comm{
+				Description: "MusicBrainz Artist Id",
+				Text:        "primary-mbid\x00collab-mbid",
+			},
+		},
+	}
+	_, ar := extractMBIDs(m)
+	if ar != "primary-mbid" {
+		t.Errorf("got %q, want %q", ar, "primary-mbid")
+	}
+}
+
+func TestCleanMBID_OnlyNULReturnsEmpty(t *testing.T) {
+	// Defensive: a TXXX value that is purely NULs (or empty after
+	// trimming) must round-trip to empty so the caller skips the row
+	// instead of writing a NUL-only string to Postgres.
+	if got := cleanMBID("\x00\x00"); got != "" {
+		t.Errorf("got %q, want empty", got)
+	}
+	if got := cleanMBID(""); got != "" {
+		t.Errorf("got %q, want empty", got)
+	}
+}