fix(diarization): correct sherpa-onnx symbol name + lint cleanup

CI failures on #9654:

* sherpa-onnx-grpc-{tts,transcription} and sherpa-onnx-realtime panicked
  at backend startup with `undefined symbol: SherpaOnnxDestroyOfflineSpeakerDiarizationResult`.
  Upstream's actual symbol is SherpaOnnxOfflineSpeakerDiarizationDestroyResult
  (Destroy in the middle, not the prefix); the rest of the diarization
  surface follows the same naming pattern. The mismatched name made
  purego.RegisterLibFunc fail at dlopen time and crashed the gRPC server
  before the BeforeAll could probe Health, taking down every sherpa-onnx
  test job — not just the diarization-related ones.

* golangci-lint flagged 5 errcheck violations on new defer cleanups
  (os.RemoveAll / Close / conn.Close); wrap each in a `defer func() { _ = X() }()`
  closure (matches the pattern other LocalAI files use for new code, since
  pre-existing bare defers are grandfathered in via new-from-merge-base).

* golangci-lint also flagged forbidigo violations: the new
  diarization_test.go files used testing.T-style `t.Errorf` / `t.Fatalf`,
  which are forbidden by the project's coding-style policy
  (.agents/coding-style.md). Convert both files to Ginkgo/Gomega
  Describe/It with Expect(...) — they get picked up by the existing
  TestBackend / TestOpenAI suites, no new suite plumbing needed.

* modernize linter: tightened the diarization segment loop to
  `for i := range int(numSegments)` (Go 1.22+ idiom).

Verified locally: golangci-lint with new-from-merge-base=origin/master
reports 0 issues across all touched packages, and the four mocked
diarization e2e specs in tests/e2e/mock_backend_test.go still pass.

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
Assisted-by: Claude:claude-opus-4-7 [Claude Code]

Author: mudler

backend/go/sherpa-onnx/backend.go

@@ -397,7 +397,7 @@ func loadSherpaLibsOnce() error {
 		{&sherpaOfflineSpeakerDiarizationResultGetNumSpeakers, "SherpaOnnxOfflineSpeakerDiarizationResultGetNumSpeakers"},
 		{&sherpaOfflineSpeakerDiarizationResultSortByStartTime, "SherpaOnnxOfflineSpeakerDiarizationResultSortByStartTime"},
 		{&sherpaOfflineSpeakerDiarizationDestroySegment, "SherpaOnnxOfflineSpeakerDiarizationDestroySegment"},
-		{&sherpaDestroyOfflineSpeakerDiarizationResult, "SherpaOnnxDestroyOfflineSpeakerDiarizationResult"},
+		{&sherpaDestroyOfflineSpeakerDiarizationResult, "SherpaOnnxOfflineSpeakerDiarizationDestroyResult"},
 	} {
 		purego.RegisterLibFunc(r.ptr, capi, r.name)
 	}
@@ -1438,7 +1438,7 @@ func (s *SherpaBackend) Diarize(req *pb.DiarizeRequest) (pb.DiarizeResponse, err
 	if err != nil {
 		return pb.DiarizeResponse{}, fmt.Errorf("failed to create temp dir: %w", err)
 	}
-	defer os.RemoveAll(dir)
+	defer func() { _ = os.RemoveAll(dir) }()
 
 	wavPath := filepath.Join(dir, "input.wav")
 	if err := utils.AudioToWav(req.Dst, wavPath); err != nil {
@@ -1487,13 +1487,13 @@ func (s *SherpaBackend) Diarize(req *pb.DiarizeRequest) (pb.DiarizeResponse, err
 	defer sherpaOfflineSpeakerDiarizationDestroySegment(segs)
 
 	out := make([]*pb.DiarizeSegment, 0, numSegments)
-	for i := int32(0); i < numSegments; i++ {
+	for i := range int(numSegments) {
 		var start, end float32
 		var spk int32
-		shimDiarizeSegmentAt(segs, i,
+		shimDiarizeSegmentAt(segs, int32(i),
 			unsafe.Pointer(&start), unsafe.Pointer(&end), unsafe.Pointer(&spk))
 		out = append(out, &pb.DiarizeSegment{
-			Id:      i,
+			Id:      int32(i),
 			Start:   start,
 			End:     end,
 			Speaker: strconv.FormatInt(int64(spk), 10),

core/backend/diarization_test.go

@@ -1,97 +1,76 @@
 package backend
 
 import (
-	"testing"
-
 	"github.com/mudler/LocalAI/pkg/grpc/proto"
+
+	. "github.com/onsi/ginkgo/v2"
+	. "github.com/onsi/gomega"
 )
 
-func TestDiarizationResultFromProto_NormalisesSpeakerLabels(t *testing.T) {
-	in := &proto.DiarizeResponse{
-		Duration: 10.5,
-		Language: "en",
-		Segments: []*proto.DiarizeSegment{
-			{Start: 0.0, End: 1.0, Speaker: "5", Text: "hi"},
-			{Start: 1.0, End: 2.0, Speaker: "2"},
-			{Start: 2.0, End: 3.5, Speaker: "5"},
-			{Start: 3.5, End: 4.0, Speaker: ""}, // empty → coerced to "0"
-		},
-	}
+var _ = Describe("diarizationResultFromProto", func() {
+	It("normalises raw backend speaker labels to SPEAKER_NN in first-seen order", func() {
+		in := &proto.DiarizeResponse{
+			Duration: 10.5,
+			Language: "en",
+			Segments: []*proto.DiarizeSegment{
+				{Start: 0.0, End: 1.0, Speaker: "5", Text: "hi"},
+				{Start: 1.0, End: 2.0, Speaker: "2"},
+				{Start: 2.0, End: 3.5, Speaker: "5"},
+				{Start: 3.5, End: 4.0, Speaker: ""}, // empty → coerced to "0"
+			},
+		}
 
-	got := diarizationResultFromProto(in)
+		got := diarizationResultFromProto(in)
 
-	if got.Task != "diarize" {
-		t.Errorf("task = %q, want diarize", got.Task)
-	}
-	if got.NumSpeakers != 3 {
-		t.Errorf("num_speakers = %d, want 3 (5, 2, 0)", got.NumSpeakers)
-	}
-	if got.Duration != 10.5 || got.Language != "en" {
-		t.Errorf("metadata not preserved: duration=%v language=%q", got.Duration, got.Language)
-	}
-	if len(got.Segments) != 4 {
-		t.Fatalf("segments=%d, want 4", len(got.Segments))
-	}
-	// First-seen-order normalization: "5"→SPEAKER_00, "2"→SPEAKER_01, ""→SPEAKER_02
-	want := []struct {
-		speaker string
-		label   string
-	}{
-		{"SPEAKER_00", "5"},
-		{"SPEAKER_01", "2"},
-		{"SPEAKER_00", "5"},
-		{"SPEAKER_02", "0"},
-	}
-	for i, w := range want {
-		if got.Segments[i].Speaker != w.speaker {
-			t.Errorf("seg[%d].speaker = %q, want %q", i, got.Segments[i].Speaker, w.speaker)
+		Expect(got.Task).To(Equal("diarize"))
+		Expect(got.NumSpeakers).To(Equal(3), "expected 3 distinct speakers (5, 2, 0)")
+		Expect(got.Duration).To(BeEquivalentTo(10.5))
+		Expect(got.Language).To(Equal("en"))
+		Expect(got.Segments).To(HaveLen(4))
+
+		// First-seen-order normalisation: "5"→SPEAKER_00, "2"→SPEAKER_01, ""→SPEAKER_02
+		want := []struct {
+			speaker string
+			label   string
+		}{
+			{"SPEAKER_00", "5"},
+			{"SPEAKER_01", "2"},
+			{"SPEAKER_00", "5"},
+			{"SPEAKER_02", "0"},
 		}
-		if got.Segments[i].Label != w.label {
-			t.Errorf("seg[%d].label = %q, want %q", i, got.Segments[i].Label, w.label)
+		for i, w := range want {
+			Expect(got.Segments[i].Speaker).To(Equal(w.speaker), "seg[%d].speaker", i)
+			Expect(got.Segments[i].Label).To(Equal(w.label), "seg[%d].label", i)
 		}
-	}
 
-	// Per-speaker totals reflect cumulative speech duration and segment count.
-	if len(got.Speakers) != 3 {
-		t.Fatalf("speakers=%d, want 3", len(got.Speakers))
-	}
-	byID := map[string]float64{}
-	countByID := map[string]int{}
-	for _, sp := range got.Speakers {
-		byID[sp.Id] = sp.TotalSpeechDuration
-		countByID[sp.Id] = sp.SegmentCount
-	}
-	if byID["SPEAKER_00"] != 2.5 { // 1.0 + 1.5
-		t.Errorf("SPEAKER_00 total = %v, want 2.5", byID["SPEAKER_00"])
-	}
-	if byID["SPEAKER_01"] != 1.0 {
-		t.Errorf("SPEAKER_01 total = %v, want 1.0", byID["SPEAKER_01"])
-	}
-	if countByID["SPEAKER_00"] != 2 || countByID["SPEAKER_01"] != 1 || countByID["SPEAKER_02"] != 1 {
-		t.Errorf("segment counts wrong: %v", countByID)
-	}
-}
+		// Per-speaker totals reflect cumulative speech duration and segment count.
+		Expect(got.Speakers).To(HaveLen(3))
+		byID := map[string]float64{}
+		countByID := map[string]int{}
+		for _, sp := range got.Speakers {
+			byID[sp.Id] = sp.TotalSpeechDuration
+			countByID[sp.Id] = sp.SegmentCount
+		}
+		Expect(byID["SPEAKER_00"]).To(BeNumerically("~", 2.5, 0.001), "1.0 + 1.5")
+		Expect(byID["SPEAKER_01"]).To(BeNumerically("~", 1.0, 0.001))
+		Expect(countByID["SPEAKER_00"]).To(Equal(2))
+		Expect(countByID["SPEAKER_01"]).To(Equal(1))
+		Expect(countByID["SPEAKER_02"]).To(Equal(1))
+	})
 
-func TestDiarizationResultFromProto_NilSafe(t *testing.T) {
-	got := diarizationResultFromProto(nil)
-	if got == nil || got.Segments == nil {
-		t.Fatalf("nil input must return a non-nil result with a non-nil segments slice")
-	}
-	if len(got.Segments) != 0 {
-		t.Errorf("nil input segments = %d, want 0", len(got.Segments))
-	}
-}
+	It("returns a non-nil result with a non-nil segments slice for nil input", func() {
+		got := diarizationResultFromProto(nil)
+		Expect(got).ToNot(BeNil())
+		Expect(got.Segments).ToNot(BeNil())
+		Expect(got.Segments).To(BeEmpty())
+	})
 
-func TestDiarizationResultFromProto_EmptySegmentsKeepsBackendSpeakerCount(t *testing.T) {
-	// When the backend reports a non-zero NumSpeakers but no segments
-	// (early stop, silence-only audio after VAD trim, etc.), the result
-	// should still surface the backend's count rather than zero.
-	in := &proto.DiarizeResponse{NumSpeakers: 2, Duration: 5}
-	got := diarizationResultFromProto(in)
-	if got.NumSpeakers != 2 {
-		t.Errorf("num_speakers = %d, want 2", got.NumSpeakers)
-	}
-	if len(got.Segments) != 0 {
-		t.Errorf("expected no segments, got %d", len(got.Segments))
-	}
-}
+	It("keeps the backend speaker count when no segments are returned", func() {
+		// Backend reports a non-zero NumSpeakers but no segments (early stop,
+		// silence-only audio after VAD trim). Surface the backend's count.
+		in := &proto.DiarizeResponse{NumSpeakers: 2, Duration: 5}
+		got := diarizationResultFromProto(in)
+		Expect(got.NumSpeakers).To(Equal(2))
+		Expect(got.Segments).To(BeEmpty())
+	})
+})

core/http/endpoints/openai/diarization.go

@@ -84,13 +84,13 @@ func DiarizationEndpoint(cl *config.ModelConfigLoader, ml *model.ModelLoader, ap
 		if err != nil {
 			return err
 		}
-		defer f.Close()
+		defer func() { _ = f.Close() }()
 
 		dir, err := os.MkdirTemp("", "diarize")
 		if err != nil {
 			return err
 		}
-		defer os.RemoveAll(dir)
+		defer func() { _ = os.RemoveAll(dir) }()
 
 		dst := filepath.Join(dir, path.Base(file.Filename))
 		dstFile, err := os.Create(dst)

core/http/endpoints/openai/diarization_test.go

@@ -2,67 +2,50 @@ package openai
 
 import (
 	"strings"
-	"testing"
 
 	"github.com/mudler/LocalAI/core/schema"
-)
-
-func TestRenderRTTM_FormatsSegmentsAsNISTRows(t *testing.T) {
-	r := &schema.DiarizationResult{
-		Segments: []schema.DiarizationSegment{
-			{Id: 0, Speaker: "SPEAKER_00", Start: 0, End: 2.34},
-			{Id: 1, Speaker: "SPEAKER_01", Start: 2.34, End: 4.10},
-		},
-	}
-	out := renderRTTM(r, "/tmp/uploads/meeting.wav")
-
-	lines := strings.Split(strings.TrimSpace(out), "\n")
-	if len(lines) != 2 {
-		t.Fatalf("expected 2 RTTM rows, got %d:\n%s", len(lines), out)
-	}
-
-	// File ID should be the basename without extension; durations are
-	// (end - start) with millisecond precision.
-	wantPrefix := "SPEAKER meeting 1 "
-	if !strings.HasPrefix(lines[0], wantPrefix) {
-		t.Errorf("row 0 prefix = %q, want %q…", lines[0][:min(len(lines[0]), len(wantPrefix))], wantPrefix)
-	}
 
-	if !strings.Contains(lines[0], " 0.000 2.340 <NA> <NA> SPEAKER_00 <NA> <NA>") {
-		t.Errorf("row 0 = %q, missing expected payload", lines[0])
-	}
-	if !strings.Contains(lines[1], " 2.340 1.760 <NA> <NA> SPEAKER_01 <NA> <NA>") {
-		t.Errorf("row 1 = %q, missing expected payload", lines[1])
-	}
-}
-
-func TestRenderRTTM_NegativeDurationClampedToZero(t *testing.T) {
-	// Backends shouldn't emit end<start, but if they do (clock skew during a
-	// long pipeline), the RTTM duration must stay non-negative.
-	r := &schema.DiarizationResult{
-		Segments: []schema.DiarizationSegment{
-			{Id: 0, Speaker: "SPEAKER_00", Start: 5, End: 4},
-		},
-	}
-	out := renderRTTM(r, "x.wav")
-	if !strings.Contains(out, " 5.000 0.000 ") {
-		t.Errorf("expected duration clamped to 0, got: %q", out)
-	}
-}
-
-func TestRenderRTTM_FallbackFileID(t *testing.T) {
-	r := &schema.DiarizationResult{
-		Segments: []schema.DiarizationSegment{{Id: 0, Speaker: "SPEAKER_00", Start: 0, End: 1}},
-	}
-	out := renderRTTM(r, "")
-	if !strings.HasPrefix(out, "SPEAKER audio 1 ") {
-		t.Errorf("expected fallback file ID 'audio', got: %q", out)
-	}
-}
+	. "github.com/onsi/ginkgo/v2"
+	. "github.com/onsi/gomega"
+)
 
-func min(a, b int) int {
-	if a < b {
-		return a
-	}
-	return b
-}
+var _ = Describe("renderRTTM", func() {
+	It("formats segments as NIST RTTM rows", func() {
+		r := &schema.DiarizationResult{
+			Segments: []schema.DiarizationSegment{
+				{Id: 0, Speaker: "SPEAKER_00", Start: 0, End: 2.34},
+				{Id: 1, Speaker: "SPEAKER_01", Start: 2.34, End: 4.10},
+			},
+		}
+		out := renderRTTM(r, "/tmp/uploads/meeting.wav")
+
+		lines := strings.Split(strings.TrimSpace(out), "\n")
+		Expect(lines).To(HaveLen(2))
+
+		// File ID should be the basename without extension; durations are
+		// (end - start) with millisecond precision.
+		Expect(lines[0]).To(HavePrefix("SPEAKER meeting 1 "))
+		Expect(lines[0]).To(ContainSubstring(" 0.000 2.340 <NA> <NA> SPEAKER_00 <NA> <NA>"))
+		Expect(lines[1]).To(ContainSubstring(" 2.340 1.760 <NA> <NA> SPEAKER_01 <NA> <NA>"))
+	})
+
+	It("clamps negative duration to zero", func() {
+		// Backends shouldn't emit end<start, but if they do (clock skew during a
+		// long pipeline), the RTTM duration must stay non-negative.
+		r := &schema.DiarizationResult{
+			Segments: []schema.DiarizationSegment{
+				{Id: 0, Speaker: "SPEAKER_00", Start: 5, End: 4},
+			},
+		}
+		out := renderRTTM(r, "x.wav")
+		Expect(out).To(ContainSubstring(" 5.000 0.000 "))
+	})
+
+	It("falls back to 'audio' when the source file name is empty", func() {
+		r := &schema.DiarizationResult{
+			Segments: []schema.DiarizationSegment{{Id: 0, Speaker: "SPEAKER_00", Start: 0, End: 1}},
+		}
+		out := renderRTTM(r, "")
+		Expect(out).To(HavePrefix("SPEAKER audio 1 "))
+	})
+})

pkg/grpc/client.go

@@ -575,7 +575,7 @@ func (c *Client) Diarize(ctx context.Context, in *pb.DiarizeRequest, opts ...grp
 	if err != nil {
 		return nil, err
 	}
-	defer conn.Close()
+	defer func() { _ = conn.Close() }()
 	client := pb.NewBackendClient(conn)
 	return client.Diarize(ctx, in, opts...)
 }

tests/e2e/mock_backend_test.go

@@ -254,7 +254,7 @@ var _ = Describe("Mock Backend E2E Tests", Label("MockBackend"), func() {
 			httpClient := &http.Client{Timeout: 30 * time.Second}
 			resp, err := httpClient.Do(req)
 			Expect(err).ToNot(HaveOccurred())
-			defer resp.Body.Close()
+			defer func() { _ = resp.Body.Close() }()
 			data, err := io.ReadAll(resp.Body)
 			Expect(err).ToNot(HaveOccurred())
 			return resp, data