feat(multimodal): 视觉分层路由（多模态检测→主/视觉/OCR 工具）+ 语音 STT→TTS 降级链（异步）

internal/runtime/multimodal/vision/vision_test.go

@@ -0,0 +1,95 @@
+package vision
+
+import (
+	"context"
+	"strings"
+	"testing"
+
+	"git.yeij.top/AskaEth/Cyrene_v2/internal/runtime/model"
+)
+
+func TestDetectMultimodal(t *testing.T) {
+	t.Parallel()
+	tests := []struct {
+		name string
+		msgs []model.Message
+		want bool
+	}{
+		{name: "text only", msgs: []model.Message{model.TextMessage(model.RoleUser, "hi")}, want: false},
+		{name: "empty", msgs: nil, want: false},
+		{name: "single image", msgs: []model.Message{model.ImageMessage(model.RoleUser, "what", "data:image/png;base64,abc")}, want: true},
+		{name: "image no text", msgs: []model.Message{model.ImageMessage(model.RoleUser, "", "https://x/y.png")}, want: true},
+		{name: "image in later turn", msgs: []model.Message{
+			model.TextMessage(model.RoleUser, "hi"),
+			model.ImageMessage(model.RoleAssistant, "", "http://x/z.png"),
+		}, want: true},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			t.Parallel()
+			if got := DetectMultimodal(tt.msgs); got != tt.want {
+				t.Errorf("DetectMultimodal() = %v, want %v", got, tt.want)
+			}
+		})
+	}
+}
+
+func TestParseImageRef(t *testing.T) {
+	t.Parallel()
+	tests := []struct {
+		name       string
+		src        string
+		wantSource ImageSource
+	}{
+		{name: "http url", src: "https://a/b.png", wantSource: SourceURL},
+		{name: "https url", src: "https://a/b.png", wantSource: SourceURL},
+		{name: "file uri", src: "file:///tmp/a.png", wantSource: SourcePath},
+		{name: "data uri", src: "data:image/png;base64,abc", wantSource: SourceBase64},
+		{name: "raw base64", src: "aGVsbG8=", wantSource: SourceBase64},
+		{name: "empty rejected", src: "", wantSource: ""},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			t.Parallel()
+			ref, err := ParseImageRef(tt.src)
+			if tt.src == "" {
+				if err == nil {
+					t.Fatal("want error for empty source, got nil")
+				}
+				return
+			}
+			if err != nil {
+				t.Fatalf("ParseImageRef error: %v", err)
+			}
+			if ref.Source != tt.wantSource {
+				t.Errorf("Source = %q, want %q", ref.Source, tt.wantSource)
+			}
+		})
+	}
+}
+
+func TestResolve_Base64(t *testing.T) {
+	t.Parallel()
+	ref, err := ParseImageRef("data:image/png;base64,aGVsbG8=")
+	if err != nil {
+		t.Fatalf("ParseImageRef error: %v", err)
+	}
+	mime, data, err := Resolve(context.Background(), ref, nil)
+	if err != nil {
+		t.Fatalf("Resolve error: %v", err)
+	}
+	if mime != "image/png" {
+		t.Errorf("mime = %q, want image/png", mime)
+	}
+	if string(data) != "hello" {
+		t.Errorf("data = %q, want hello", data)
+	}
+}
+
+func TestResolve_DataURI(t *testing.T) {
+	t.Parallel()
+	got := DataURI("image/webp", []byte{1, 2, 3})
+	if !strings.HasPrefix(got, "data:image/webp;base64,") {
+		t.Errorf("DataURI = %q, want data:image/webp;base64,", got)
+	}
+}