serverless-voice-to-text-bot/main.go

package main

import (
	"context"
	"encoding/json"
	"fmt"
	"io"
	"log"
	"mime/multipart"
	"net/http"
	"os"
	"slices"
	"strconv"
	"strings"

	"github.com/carlmjohnson/requests"
)

type TelegramMessage struct {
	MessageID int            `json:"message_id,omitempty"`
	From      TelegramUser   `json:"from,omitempty"`
	Chat      TelegramChat   `json:"chat,omitempty"`
	Voice     *TelegramVoice `json:"voice,omitempty"`
}

type TelegramUser struct {
	ID       int    `json:"id,omitempty"`
	Username string `json:"username,omitempty"`
}

type TelegramChat struct {
	ID int `json:"id,omitempty"`
}

type TelegramVoice struct {
	FileID string `json:"file_id,omitempty"`
}

type TelegramSendMessage struct {
	ChatID          int                      `json:"chat_id,omitempty"`
	Text            string                   `json:"text,omitempty"`
	ReplyParameters *TelegramReplyParameters `json:"reply_parameters,omitempty"`
}

type TelegramReplyParameters struct {
	MessageID int `json:"message_id,omitempty"`
}

func main() {
	// Retrieve bot token from environment variable
	botToken := os.Getenv("BOT_TOKEN")
	if botToken == "" {
		log.Fatal("BOT_TOKEN environment variable is not set")
	}

	// Retrieve OpenAI API key
	openaiApiKey := os.Getenv("OPENAI_KEY")
	if openaiApiKey == "" {
		log.Fatal("OPENAI_KEY environment variable is not set")
	}

	// Retrieve allowed users
	allowedUsers := strings.Split(os.Getenv("ALLOWED_USERS"), ",")
	if len(allowedUsers) == 0 {
		log.Fatal("ALLOWED_USERS environment variable is not set")
	}
	log.Println("Allowed users:", allowedUsers)

	// Set up the HTTP server
	http.HandleFunc("/", func(w http.ResponseWriter, r *http.Request) {
		io.WriteString(w, "Hi!")
	})
	http.HandleFunc("/webhook", func(w http.ResponseWriter, r *http.Request) {
		// Decode the incoming message
		var update struct {
			Message *TelegramMessage `json:"message"`
		}
		if err := json.NewDecoder(r.Body).Decode(&update); err != nil {
			log.Println("Error decoding update:", err)
			return
		}
		if update.Message == nil {
			return
		}

		// Check allowed users
		if !(slices.Contains(allowedUsers, update.Message.From.Username) || slices.Contains(allowedUsers, strconv.Itoa(update.Message.From.ID))) {
			sendMessage(botToken, update.Message.Chat.ID, "Sorry, you are not allowed to use this bot!", update.Message.MessageID)
			return
		}

		if update.Message.Voice != nil {
			transcript, err := transcriptVoiceMessage(botToken, openaiApiKey, update.Message.Voice.FileID)
			if err != nil {
				log.Println("Error transcribing voice message:", err)
				sendMessage(botToken, update.Message.Chat.ID, "Failed to transcribe voice message", update.Message.MessageID)
			} else {
				for _, transcriptPart := range splitText(transcript, 3000) {
					sendMessage(botToken, update.Message.Chat.ID, transcriptPart, update.Message.MessageID)
				}
			}
		} else {
			sendMessage(botToken, update.Message.Chat.ID, "Please send a voice message", update.Message.MessageID)
		}
	})

	// Start the HTTP server
	port := os.Getenv("PORT")
	if port == "" {
		port = "8080"
	}
	log.Printf("Starting server on :%s", port)
	log.Fatal(http.ListenAndServe(":"+port, nil))
}

func sendMessage(botToken string, chatID int, text string, replyTo int) {
	msg := &TelegramSendMessage{
		ChatID: chatID,
		Text:   text,
	}
	if replyTo != 0 {
		msg.ReplyParameters = &TelegramReplyParameters{
			MessageID: replyTo,
		}
	}
	err := requests.URL(fmt.Sprintf("https://api.telegram.org/bot%s/sendMessage", botToken)).
		Method(http.MethodPost).
		BodyJSON(msg).
		Fetch(context.Background())
	if err != nil {
		log.Println("Error sending message:", err)
		return
	}
}

func transcriptVoiceMessage(botToken, openaiApiKey, fileID string) (string, error) {
	// Get file path
	var fileResponse struct {
		OK     bool `json:"ok"`
		Result struct {
			FilePath string `json:"file_path"`
		} `json:"result"`
	}
	err := requests.URL(fmt.Sprintf("https://api.telegram.org/bot%s/getFile", botToken)).
		Param("file_id", fileID).
		ToJSON(&fileResponse).
		Fetch(context.Background())
	if err != nil {
		return "", err
	}

	// Download the file
	voiceDownload := fmt.Sprintf("https://api.telegram.org/file/bot%s/%s", botToken, fileResponse.Result.FilePath)
	voiceReader, voiceWriter := io.Pipe()
	go func() {
		voiceWriter.CloseWithError(
			requests.URL(voiceDownload).
				ToWriter(voiceWriter).
				Fetch(context.Background()),
		)
	}()

	// Transcribe message
	multipartReader, multipartWriter := io.Pipe()
	mpw := multipart.NewWriter(multipartWriter)
	go func() {
		defer multipartWriter.Close()
		mpw.WriteField("model", "whisper-1")
		fileWriter, err := mpw.CreateFormFile("file", "file.oga")
		if err != nil {
			multipartWriter.CloseWithError(err)
			return
		}
		_, err = io.Copy(fileWriter, voiceReader)
		if err != nil {
			multipartWriter.CloseWithError(err)
			return
		}
		multipartWriter.CloseWithError(mpw.Close())
	}()

	var transcription struct {
		Text string `json:"text"`
	}
	err = requests.URL("https://api.openai.com/v1/audio/transcriptions").
		Method(http.MethodPost).
		Header("Authorization", fmt.Sprintf("Bearer %s", openaiApiKey)).
		ContentType(mpw.FormDataContentType()).
		BodyReader(multipartReader).
		ToJSON(&transcription).
		Fetch(context.Background())

	return transcription.Text, err
}

func splitText(s string, chunkSize int) []string {
	if chunkSize >= len(s) {
		return []string{s}
	}
	var chunks []string
	var b strings.Builder
	b.Grow(chunkSize)
	l := 0
	for _, r := range s {
		b.WriteRune(r)
		l++
		if l == chunkSize {
			chunks = append(chunks, b.String())
			l = 0
			b.Reset()
			b.Grow(chunkSize)
		}
	}
	if l > 0 {
		chunks = append(chunks, b.String())
	}
	return chunks
}