feat(canvas): add visual annotation edit action

Add a VisualAnnotationPanel letting users mark regions on an image (brush, circle, rectangle, cross) with per-mark edit instructions, composited into an annotation guide image. The new "visual-annotation" node action sends the unmarked source plus the annotated guide to the image model, editing only numbered regions. Server validates the annotation image/instructions and threads them through the generator task. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-11 10:03:59 +08:00
parent 114198f1d9
commit bb8946b79b
17 changed files with 1271 additions and 46 deletions
@@ -0,0 +1,98 @@
+package application
+
+import (
+	"context"
+	"errors"
+	"strings"
+	"testing"
+	"time"
+
+	"img_infinite_canvas/internal/domain/design"
+	"img_infinite_canvas/internal/infrastructure/memory"
+)
+
+func TestRunVisualAnnotationCarriesCleanSourceAndGuideImage(t *testing.T) {
+	ctx := context.Background()
+	repo := memory.NewProjectRepository()
+	queue := &fakeJobQueue{}
+	service := NewDesignService(repo, nil, nil, nil, nil)
+	service.SetJobQueue(queue)
+
+	project := design.Project{
+		ID:        "project-visual-annotation",
+		Title:     "Annotated edit",
+		Status:    design.StatusReady,
+		UpdatedAt: time.Now(),
+		Nodes: []design.Node{{
+			ID:      "image-1",
+			Type:    design.NodeTypeImage,
+			Title:   "Source",
+			Width:   640,
+			Height:  480,
+			Content: "https://example.com/source.png",
+			Status:  "success",
+		}},
+	}
+	if err := repo.Save(ctx, project); err != nil {
+		t.Fatal(err)
+	}
+
+	_, err := service.RunNodeActionAsync(ctx, project.ID, "image-1", design.NodeActionRequest{
+		Action: "visual-annotation",
+		Prompt: "1. Remove the logo",
+	})
+	if !errors.Is(err, design.ErrInvalidInput) {
+		t.Fatalf("expected missing guide image to be rejected, got %v", err)
+	}
+
+	selection := `{"annotations":[{"index":1,"type":"cross","instruction":"Remove the logo","points":[{"x":0.4,"y":0.3},{"x":0.6,"y":0.5}]}]}`
+	guideURL := "https://example.com/annotation-guide.png"
+	next, err := service.RunNodeActionAsync(ctx, project.ID, "image-1", design.NodeActionRequest{
+		Action:          "visual_annotation",
+		Prompt:          "1. Cross: Remove the logo",
+		Selection:       selection,
+		AnnotationImage: guideURL,
+	})
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	if len(next.Nodes) != 2 {
+		t.Fatalf("expected the source plus an annotated-edit copy, got %#v", next.Nodes)
+	}
+	sourceNode := next.Nodes[0]
+	placeholder := next.Nodes[1]
+	if sourceNode.Status != "success" || sourceNode.Title != "Source" || sourceNode.Content != project.Nodes[0].Content {
+		t.Fatalf("expected the source image to remain unchanged, got %#v", sourceNode)
+	}
+	if placeholder.Status != "generating" || placeholder.Title != "标注改图中" || placeholder.ParentID != sourceNode.ID || placeholder.LayerRole != "visual-annotation-edit" {
+		t.Fatalf("expected a generating annotated-edit copy, got %#v", placeholder)
+	}
+	if placeholder.X != sourceNode.X+sourceNode.Width+32 || placeholder.Y != sourceNode.Y || placeholder.Content != sourceNode.Content {
+		t.Fatalf("expected the annotated-edit copy beside the source, got %#v", placeholder)
+	}
+	if len(queue.jobs) != 1 {
+		t.Fatalf("expected one visual annotation job, got %#v", queue.jobs)
+	}
+	request := queue.jobs[0].Request
+	if request.Action != "visual-annotation" || request.AnnotationImage != guideURL || request.Selection != selection {
+		t.Fatalf("visual annotation request was not preserved in the job: %#v", request)
+	}
+	if request.Options[nodeActionResultNodeIDOption] != placeholder.ID {
+		t.Fatalf("expected the job to target copied node %q, got %#v", placeholder.ID, request.Options)
+	}
+
+	task, err := service.GeneratorTask(ctx, project.ID, next.LastThreadID)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(task.OriginalInputArgs.Image) != 2 || task.OriginalInputArgs.Image[0] != project.Nodes[0].Content || task.OriginalInputArgs.Image[1] != guideURL {
+		t.Fatalf("expected clean source followed by annotation guide, got %#v", task.OriginalInputArgs.Image)
+	}
+	prompt := buildNodeActionPrompt(request, project.Nodes[0])
+	for _, fragment := range []string{"first image is the clean source", "second image is an annotated guide only", "never reproduce guide marks", selection} {
+		if !strings.Contains(prompt, fragment) {
+			t.Fatalf("expected visual annotation prompt to contain %q, got %q", fragment, prompt)
+		}
+	}
+}