From 59970efd41b3bddc4afe5d50c1f7b2b51e972fa5 Mon Sep 17 00:00:00 2001
From: linhaolin1 <linhaolin1@gmail.com>
Date: Thu, 2 Apr 2026 23:29:46 +0800
Subject: [PATCH] feat(vk): add voice capabilities support

- Implement VoiceCapabilities() method for VK channel
- Add audio_message attachment handling in processAttachments
- Add comprehensive tests for voice capabilities
- Support both ASR (speech-to-text) and TTS (text-to-speech)
---
 pkg/channels/vk/vk.go      |  6 ++++++
 pkg/channels/vk/vk_test.go | 36 +++++++++++++++++++++++++++++++++++-
 2 files changed, 41 insertions(+), 1 deletion(-)

diff --git a/pkg/channels/vk/vk.go b/pkg/channels/vk/vk.go
index 0bc373d18..92fbcf4ad 100644
--- a/pkg/channels/vk/vk.go
+++ b/pkg/channels/vk/vk.go
@@ -271,6 +271,8 @@ func (c *VKChannel) processAttachments(attachments []object.MessagesMessageAttac
 			} else {
 				parts = append(parts, "[document]")
 			}
+		case "audio_message":
+			parts = append(parts, "[voice]")
 		case "sticker":
 			parts = append(parts, "[sticker]")
 		}
@@ -278,3 +280,7 @@ func (c *VKChannel) processAttachments(attachments []object.MessagesMessageAttac
 
 	return strings.Join(parts, " ")
 }
+
+func (c *VKChannel) VoiceCapabilities() channels.VoiceCapabilities {
+	return channels.VoiceCapabilities{ASR: true, TTS: true}
+}
diff --git a/pkg/channels/vk/vk_test.go b/pkg/channels/vk/vk_test.go
index e4fdc1ac3..c7e62ab31 100644
--- a/pkg/channels/vk/vk_test.go
+++ b/pkg/channels/vk/vk_test.go
@@ -202,6 +202,11 @@ func TestVKChannel_ProcessAttachments(t *testing.T) {
 			attachments: []string{"sticker"},
 			want:        "[sticker]",
 		},
+		{
+			name:        "audio_message attachment",
+			attachments: []string{"audio_message"},
+			want:        "[voice]",
+		},
 		{
 			name:        "multiple attachments",
 			attachments: []string{"photo", "video", "audio"},
@@ -216,7 +221,11 @@ func TestVKChannel_ProcessAttachments(t *testing.T) {
 				if i > 0 {
 					result += " "
 				}
-				result += "[" + att + "]"
+				if att == "audio_message" {
+					result += "[voice]"
+				} else {
+					result += "[" + att + "]"
+				}
 			}
 			if result != tt.want {
 				t.Errorf("processAttachments() = %q, want %q", result, tt.want)
@@ -224,3 +233,28 @@ func TestVKChannel_ProcessAttachments(t *testing.T) {
 		})
 	}
 }
+
+func TestVKChannel_VoiceCapabilities(t *testing.T) {
+	msgBus := bus.NewMessageBus()
+	cfg := &config.Config{
+		Channels: config.ChannelsConfig{
+			VK: config.VKConfig{
+				Enabled: true,
+				Token:   *config.NewSecureString("test_token"),
+				GroupID: 123456789,
+			},
+		},
+	}
+	ch, err := NewVKChannel(cfg, msgBus)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+
+	caps := ch.VoiceCapabilities()
+	if !caps.ASR {
+		t.Error("VoiceCapabilities().ASR should be true")
+	}
+	if !caps.TTS {
+		t.Error("VoiceCapabilities().TTS should be true")
+	}
+}