feat(telegram): include quoted reply context and media in inbound turns (#2200)

2026-06-12 18:08:54 +00:00 · 2026-03-31 04:46:41 +02:00
parent 073cc3f65e
commit 2d8556205f
2 changed files with 319 additions and 0 deletions
@@ -660,6 +660,23 @@ func (c *TelegramChannel) handleMessage(ctx context.Context, message *telego.Mes
 		content = cleaned
 	}

+	if message.ReplyToMessage != nil {
+		quotedMedia := quotedTelegramMediaRefs(
+			message.ReplyToMessage,
+			func(fileID, ext, filename string) string {
+				localPath := c.downloadFile(ctx, fileID, ext)
+				if localPath == "" {
+					return ""
+				}
+				return storeMedia(localPath, filename)
+			},
+		)
+		if len(quotedMedia) > 0 {
+			mediaPaths = append(quotedMedia, mediaPaths...)
+		}
+		content = c.prependTelegramQuotedReply(content, message.ReplyToMessage)
+	}
+
 	// For forum topics, embed the thread ID as "chatID/threadID" so replies
 	// route to the correct topic and each topic gets its own session.
 	// Only forum groups (IsForum) are handled; regular group reply threads
@@ -693,6 +710,9 @@ func (c *TelegramChannel) handleMessage(ctx context.Context, message *telego.Mes
 		"first_name": user.FirstName,
 		"is_group":   fmt.Sprintf("%t", message.Chat.Type != "private"),
 	}
+	if message.ReplyToMessage != nil {
+		metadata["reply_to_message_id"] = fmt.Sprintf("%d", message.ReplyToMessage.MessageID)
+	}

 	// Set parent_peer metadata for per-topic agent binding.
 	if message.Chat.IsForum && threadID != 0 {
@@ -713,6 +733,122 @@ func (c *TelegramChannel) handleMessage(ctx context.Context, message *telego.Mes
 	return nil
 }

+func (c *TelegramChannel) prependTelegramQuotedReply(content string, reply *telego.Message) string {
+	quoted := strings.TrimSpace(telegramQuotedContent(reply))
+	if quoted == "" {
+		return content
+	}
+
+	author := telegramQuotedAuthor(reply)
+	role := c.telegramQuotedRole(reply)
+	if strings.TrimSpace(content) == "" {
+		return fmt.Sprintf("[quoted %s message from %s]: %s", role, author, quoted)
+	}
+	return fmt.Sprintf("[quoted %s message from %s]: %s\n\n%s", role, author, quoted, content)
+}
+
+func (c *TelegramChannel) telegramQuotedRole(message *telego.Message) string {
+	if message == nil {
+		return "unknown"
+	}
+
+	if message.From != nil {
+		if !message.From.IsBot {
+			return "user"
+		}
+		if c.isOwnBotUser(message.From) {
+			return "assistant"
+		}
+		return "bot"
+	}
+
+	if message.SenderChat != nil {
+		return "chat"
+	}
+
+	return "unknown"
+}
+
+func (c *TelegramChannel) isOwnBotUser(user *telego.User) bool {
+	if c == nil || c.bot == nil || user == nil || !user.IsBot {
+		return false
+	}
+
+	if botID := c.bot.ID(); botID != 0 && user.ID == botID {
+		return true
+	}
+
+	botUsername := strings.TrimPrefix(strings.TrimSpace(c.bot.Username()), "@")
+	if botUsername == "" {
+		return false
+	}
+	return strings.EqualFold(strings.TrimPrefix(strings.TrimSpace(user.Username), "@"), botUsername)
+}
+
+func telegramQuotedAuthor(message *telego.Message) string {
+	if message == nil || message.From == nil {
+		return "unknown"
+	}
+	if username := strings.TrimSpace(message.From.Username); username != "" {
+		return username
+	}
+	if firstName := strings.TrimSpace(message.From.FirstName); firstName != "" {
+		return firstName
+	}
+	return "unknown"
+}
+
+func telegramQuotedContent(message *telego.Message) string {
+	if message == nil {
+		return ""
+	}
+
+	var parts []string
+	if text := strings.TrimSpace(message.Text); text != "" {
+		parts = append(parts, text)
+	}
+	if caption := strings.TrimSpace(message.Caption); caption != "" {
+		parts = append(parts, caption)
+	}
+	switch {
+	case len(message.Photo) > 0:
+		parts = append(parts, "[image: photo]")
+	}
+	switch {
+	case message.Voice != nil:
+		parts = append(parts, "[voice]")
+	case message.Audio != nil:
+		parts = append(parts, "[audio]")
+	}
+	if message.Document != nil {
+		parts = append(parts, "[file]")
+	}
+
+	return strings.Join(parts, "\n")
+}
+
+func quotedTelegramMediaRefs(
+	message *telego.Message,
+	resolve func(fileID, ext, filename string) string,
+) []string {
+	if message == nil || resolve == nil {
+		return nil
+	}
+
+	var refs []string
+	if message.Voice != nil {
+		if ref := resolve(message.Voice.FileID, ".ogg", "voice.ogg"); ref != "" {
+			refs = append(refs, ref)
+		}
+	}
+	if message.Audio != nil {
+		if ref := resolve(message.Audio.FileID, ".mp3", "audio.mp3"); ref != "" {
+			refs = append(refs, ref)
+		}
+	}
+	return refs
+}
+
 func (c *TelegramChannel) downloadPhoto(ctx context.Context, fileID string) string {
 	file, err := c.bot.GetFile(ctx, &telego.GetFileParams{FileID: fileID})
 	if err != nil {
@@ -7,6 +7,7 @@ import (
 	"io"
 	"os"
 	"path/filepath"
+	"strconv"
 	"strings"
 	"testing"

@@ -104,6 +105,13 @@ func successResponse(t *testing.T) *ta.Response {
 	return &ta.Response{Ok: true, Result: b}
 }

+func successUserResponse(t *testing.T, user *telego.User) *ta.Response {
+	t.Helper()
+	b, err := json.Marshal(user)
+	require.NoError(t, err)
+	return &ta.Response{Ok: true, Result: b}
+}
+
 // newTestChannel creates a TelegramChannel with a mocked bot for unit testing.
 func newTestChannel(t *testing.T, caller *stubCaller) *TelegramChannel {
 	return newTestChannelWithConstructor(t, caller, &stubConstructor{})
@@ -642,6 +650,181 @@ func TestHandleMessage_ReplyThread_NonForum_NoIsolation(t *testing.T) {
 	assert.Empty(t, inbound.Metadata["parent_peer_id"])
 }

+func assertHandleMessageQuotedUserReply(
+	t *testing.T,
+	chatID int64,
+	messageID int,
+	userID int64,
+	userName string,
+	userText string,
+	replyMessageID int,
+	replyText string,
+	replyCaption string,
+	replyAuthorID int64,
+	replyAuthorName string,
+	expectedContent string,
+) {
+	t.Helper()
+
+	messageBus := bus.NewMessageBus()
+	ch := &TelegramChannel{
+		BaseChannel: channels.NewBaseChannel("telegram", nil, messageBus, nil),
+		chatIDs:     make(map[string]int64),
+		ctx:         context.Background(),
+	}
+
+	msg := &telego.Message{
+		Text:      userText,
+		MessageID: messageID,
+		Chat: telego.Chat{
+			ID:   chatID,
+			Type: "private",
+		},
+		From: &telego.User{
+			ID:        userID,
+			FirstName: userName,
+		},
+		ReplyToMessage: &telego.Message{
+			MessageID: replyMessageID,
+			Text:      replyText,
+			Caption:   replyCaption,
+			From: &telego.User{
+				ID:        replyAuthorID,
+				FirstName: replyAuthorName,
+			},
+		},
+	}
+
+	err := ch.handleMessage(context.Background(), msg)
+	require.NoError(t, err)
+
+	inbound, ok := <-messageBus.InboundChan()
+	require.True(t, ok)
+	assert.Equal(t, strconv.Itoa(replyMessageID), inbound.Metadata["reply_to_message_id"])
+	assert.Equal(t, expectedContent, inbound.Content)
+}
+
+func TestHandleMessage_ReplyToMessage_PrependsQuotedTextAndMetadata(t *testing.T) {
+	assertHandleMessageQuotedUserReply(
+		t,
+		456,
+		21,
+		11,
+		"Alice",
+		"follow up",
+		99,
+		"old context",
+		"",
+		12,
+		"Bob",
+		"[quoted user message from Bob]: old context\n\nfollow up",
+	)
+}
+
+func TestHandleMessage_ReplyToMessage_UsesCaptionWhenQuotedTextMissing(t *testing.T) {
+	assertHandleMessageQuotedUserReply(
+		t,
+		789,
+		22,
+		13,
+		"Carol",
+		"answer this",
+		100,
+		"",
+		"caption context",
+		14,
+		"Dave",
+		"[quoted user message from Dave]: caption context\n\nanswer this",
+	)
+}
+
+func TestHandleMessage_ReplyToOwnBotMessage_UsesAssistantRole(t *testing.T) {
+	messageBus := bus.NewMessageBus()
+	caller := &stubCaller{
+		callFn: func(ctx context.Context, url string, data *ta.RequestData) (*ta.Response, error) {
+			if strings.Contains(url, "getMe") {
+				return successUserResponse(t, &telego.User{
+					ID:        42,
+					IsBot:     true,
+					FirstName: "Pico",
+					Username:  "afjcjsbx_picoclaw_bot",
+				}), nil
+			}
+			t.Fatalf("unexpected API call: %s", url)
+			return nil, nil
+		},
+	}
+	ch := newTestChannel(t, caller)
+	ch.BaseChannel = channels.NewBaseChannel("telegram", nil, messageBus, nil)
+	ch.ctx = context.Background()
+
+	msg := &telego.Message{
+		Text:      "ti ricordi questo file?",
+		MessageID: 23,
+		Chat: telego.Chat{
+			ID:   999,
+			Type: "private",
+		},
+		From: &telego.User{
+			ID:        15,
+			FirstName: "Eve",
+		},
+		ReplyToMessage: &telego.Message{
+			MessageID: 101,
+			Text:      "Fatto! Ho creato il file notizie_2026_03_28.md",
+			From: &telego.User{
+				ID:        42,
+				IsBot:     true,
+				FirstName: "Pico",
+				Username:  "afjcjsbx_picoclaw_bot",
+			},
+		},
+	}
+
+	err := ch.handleMessage(context.Background(), msg)
+	require.NoError(t, err)
+
+	inbound, ok := <-messageBus.InboundChan()
+	require.True(t, ok)
+	assert.Equal(t, "101", inbound.Metadata["reply_to_message_id"])
+	assert.Equal(
+		t,
+		"[quoted assistant message from afjcjsbx_picoclaw_bot]: Fatto! Ho creato il file notizie_2026_03_28.md\n\nti ricordi questo file?",
+		inbound.Content,
+	)
+}
+
+func TestTelegramQuotedContent_IncludesVoiceMarkerAlongsideCaption(t *testing.T) {
+	msg := &telego.Message{
+		Caption: "listen to this",
+		Voice: &telego.Voice{
+			FileID: "voice-file",
+		},
+	}
+
+	assert.Equal(t, "listen to this\n[voice]", telegramQuotedContent(msg))
+}
+
+func TestQuotedTelegramMediaRefs_ResolvesQuotedAudioInOrder(t *testing.T) {
+	msg := &telego.Message{
+		Voice: &telego.Voice{FileID: "voice-file"},
+		Audio: &telego.Audio{FileID: "audio-file"},
+	}
+
+	var calls []string
+	refs := quotedTelegramMediaRefs(msg, func(fileID, ext, filename string) string {
+		calls = append(calls, fileID+"|"+ext+"|"+filename)
+		return "ref://" + filename
+	})
+
+	assert.Equal(
+		t,
+		[]string{"voice-file|.ogg|voice.ogg", "audio-file|.mp3|audio.mp3"},
+		calls,
+	)
+	assert.Equal(t, []string{"ref://voice.ogg", "ref://audio.mp3"}, refs)
+}
+
 func TestHandleMessage_EmptyContent_Ignored(t *testing.T) {
 	messageBus := bus.NewMessageBus()
 	ch := &TelegramChannel{