feat: persist voice samples to disk, generate VTT, cleanup after analysis

internal/infrastructure/speech/openai_whisper.go

@@ -8,7 +8,10 @@ import (
 	"io"
 	"mime/multipart"
 	"net/http"
+	"strings"
 	"time"
+
+	"github.com/paramah/gw_telegram/internal/domain/port"
 )
 
 const whisperAPIURL = "https://api.openai.com/v1/audio/transcriptions"
@@ -29,61 +32,97 @@ func NewOpenAIWhisper(apiKey, model, language string) *OpenAIWhisper {
 	}
 }
 
-type whisperResponse struct {
-	Text string `json:"text"`
+// whisperVerboseResponse is returned by Whisper API when response_format=verbose_json.
+type whisperVerboseResponse struct {
+	Text     string           `json:"text"`
+	Segments []whisperSegment `json:"segments"`
 }
 
-func (w *OpenAIWhisper) Transcribe(ctx context.Context, audioData []byte, mimeType string) (string, error) {
+type whisperSegment struct {
+	Start float64 `json:"start"`
+	End   float64 `json:"end"`
+	Text  string  `json:"text"`
+}
+
+func (w *OpenAIWhisper) Transcribe(ctx context.Context, audioData []byte, mimeType string) (port.TranscriptionResult, error) {
 	var buf bytes.Buffer
 	mw := multipart.NewWriter(&buf)
 
 	fw, err := mw.CreateFormFile("file", "audio.wav")
 	if err != nil {
-		return "", fmt.Errorf("create form file: %w", err)
+		return port.TranscriptionResult{}, fmt.Errorf("create form file: %w", err)
 	}
 	if _, err := fw.Write(audioData); err != nil {
-		return "", fmt.Errorf("write audio data: %w", err)
+		return port.TranscriptionResult{}, fmt.Errorf("write audio data: %w", err)
 	}
-
 	if err := mw.WriteField("model", w.model); err != nil {
-		return "", fmt.Errorf("write model field: %w", err)
+		return port.TranscriptionResult{}, fmt.Errorf("write model field: %w", err)
 	}
-	if err := mw.WriteField("response_format", "json"); err != nil {
-		return "", fmt.Errorf("write response_format: %w", err)
+	if err := mw.WriteField("response_format", "verbose_json"); err != nil {
+		return port.TranscriptionResult{}, fmt.Errorf("write response_format: %w", err)
 	}
 	if w.language != "" {
 		if err := mw.WriteField("language", w.language); err != nil {
-			return "", fmt.Errorf("write language field: %w", err)
+			return port.TranscriptionResult{}, fmt.Errorf("write language field: %w", err)
 		}
 	}
 	mw.Close()
 
 	req, err := http.NewRequestWithContext(ctx, http.MethodPost, whisperAPIURL, &buf)
 	if err != nil {
-		return "", fmt.Errorf("create whisper request: %w", err)
+		return port.TranscriptionResult{}, fmt.Errorf("create whisper request: %w", err)
 	}
 	req.Header.Set("Authorization", "Bearer "+w.apiKey)
 	req.Header.Set("Content-Type", mw.FormDataContentType())
 
 	resp, err := w.client.Do(req)
 	if err != nil {
-		return "", fmt.Errorf("whisper API call: %w", err)
+		return port.TranscriptionResult{}, fmt.Errorf("whisper API call: %w", err)
 	}
 	defer resp.Body.Close()
 
 	body, err := io.ReadAll(resp.Body)
 	if err != nil {
-		return "", fmt.Errorf("read whisper response: %w", err)
+		return port.TranscriptionResult{}, fmt.Errorf("read whisper response: %w", err)
 	}
-
 	if resp.StatusCode != http.StatusOK {
-		return "", fmt.Errorf("whisper API error %d: %s", resp.StatusCode, string(body))
+		return port.TranscriptionResult{}, fmt.Errorf("whisper API error %d: %s", resp.StatusCode, string(body))
 	}
 
-	var result whisperResponse
+	var result whisperVerboseResponse
 	if err := json.Unmarshal(body, &result); err != nil {
-		return "", fmt.Errorf("parse whisper response: %w", err)
+		return port.TranscriptionResult{}, fmt.Errorf("parse whisper response: %w", err)
 	}
 
-	return result.Text, nil
+	return port.TranscriptionResult{
+		Text: result.Text,
+		VTT:  buildVTT(result.Segments),
+	}, nil
+}
+
+// buildVTT generates WebVTT content from Whisper segments.
+func buildVTT(segments []whisperSegment) string {
+	if len(segments) == 0 {
+		return ""
+	}
+	var sb strings.Builder
+	sb.WriteString("WEBVTT\n\n")
+	for _, seg := range segments {
+		sb.WriteString(fmt.Sprintf("%s --> %s\n%s\n\n",
+			formatVTTTime(seg.Start),
+			formatVTTTime(seg.End),
+			strings.TrimSpace(seg.Text),
+		))
+	}
+	return sb.String()
+}
+
+// formatVTTTime converts seconds to WebVTT timestamp (HH:MM:SS.mmm).
+func formatVTTTime(seconds float64) string {
+	d := time.Duration(seconds * float64(time.Second))
+	h := int(d.Hours())
+	m := int(d.Minutes()) % 60
+	s := int(d.Seconds()) % 60
+	ms := int(d.Milliseconds()) % 1000
+	return fmt.Sprintf("%02d:%02d:%02d.%03d", h, m, s, ms)
 }