Files
ai-service/internal/transcription/client_test.go
Grendgi e6c2b46cf6
All checks were successful
CI / test (push) Successful in 14s
Build and Deploy / build-and-deploy (push) Successful in 30s
Use Voxtral audio transcription endpoint
2026-06-09 15:51:50 +03:00

118 lines
3.3 KiB
Go

package transcription
import (
"encoding/json"
"math"
"net/http"
"net/http/httptest"
"testing"
"time"
)
func TestAdjustLeadSilence(t *testing.T) {
got := adjustLeadSilence([]Segment{
{Start: 0.2, End: 1.1, Text: "first"},
{Start: 1.4, End: 2.0, Text: "second"},
}, 800*time.Millisecond)
if got[0].Start != 0 {
t.Fatalf("first start = %v, want 0", got[0].Start)
}
if !near(got[0].End, 0.3) {
t.Fatalf("first end = %v, want 0.3", got[0].End)
}
if !near(got[1].Start, 0.6) {
t.Fatalf("second start = %v, want 0.6", got[1].Start)
}
}
func TestNormalizeProviderOrder(t *testing.T) {
got := normalizeProviderOrder([]string{"whisperx", "qwen", "voxtral", "qwen2-audio"})
want := []string{ProviderWhisperX, ProviderQwenAudio, ProviderVoxtral}
if len(got) != len(want) {
t.Fatalf("providers = %#v, want %#v", got, want)
}
for i := range want {
if got[i] != want[i] {
t.Fatalf("providers = %#v, want %#v", got, want)
}
}
}
func TestNewWithOptionsBuildsComparisonProviders(t *testing.T) {
client := NewWithOptions(Options{
Providers: []string{"whisperx", "qwen2-audio", "voxtral-small"},
WhisperXURL: "http://whisperx",
QwenAudioBaseURL: "http://qwen",
VoxtralBaseURL: "http://voxtral",
})
if client == nil {
t.Fatal("client is nil")
}
got := make([]string, 0, len(client.providers))
for _, provider := range client.providers {
got = append(got, provider.Name)
}
want := []string{ProviderWhisperX, ProviderQwenAudio, ProviderVoxtral}
for i := range want {
if got[i] != want[i] {
t.Fatalf("providers = %#v, want %#v", got, want)
}
}
}
func TestAudioDataURLUsesVLLMAudioURLFormat(t *testing.T) {
got := audioDataURL([]byte("abc"), "call.wav")
want := "data:audio/wav;base64,YWJj"
if got != want {
t.Fatalf("audio data url = %q, want %q", got, want)
}
}
func TestVoxtralUsesAudioTranscriptionsEndpoint(t *testing.T) {
audioSrv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
_, _ = w.Write([]byte("fake audio"))
}))
defer audioSrv.Close()
var gotPath, gotModel string
providerSrv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
gotPath = r.URL.Path
if err := r.ParseMultipartForm(16 << 20); err != nil {
t.Fatalf("ParseMultipartForm: %v", err)
}
gotModel = r.FormValue("model")
if _, _, err := r.FormFile("file"); err != nil {
t.Fatalf("FormFile: %v", err)
}
_ = json.NewEncoder(w).Encode(map[string]string{"text": "Алло, тест."})
}))
defer providerSrv.Close()
client := NewWithOptions(Options{
Providers: []string{"voxtral-small"},
VoxtralBaseURL: providerSrv.URL,
VoxtralModel: "mistralai/Voxtral-Small-24B-2507",
})
if client == nil {
t.Fatal("client is nil")
}
got, err := client.Transcribe(t.Context(), Input{AudioURL: audioSrv.URL, Filename: "call.mp3"})
if err != nil {
t.Fatalf("Transcribe: %v", err)
}
if gotPath != "/v1/audio/transcriptions" {
t.Fatalf("path = %q, want /v1/audio/transcriptions", gotPath)
}
if gotModel != "mistralai/Voxtral-Small-24B-2507" {
t.Fatalf("model = %q", gotModel)
}
if len(got.Segments) != 1 || got.Segments[0].Text != "Алло, тест." {
t.Fatalf("segments = %#v", got.Segments)
}
}
func near(got, want float64) bool {
return math.Abs(got-want) < 0.000001
}