feat(seedance): 同步输入图片约束与自动转换

2026-07-23 21:55:42 +08:00
parent 46cdb1a288
commit 8b7d3e9c9a
7 changed files with 656 additions and 0 deletions
@@ -0,0 +1,140 @@
+package runner
+
+import (
+	"bytes"
+	"context"
+	"encoding/base64"
+	"image"
+	"image/color"
+	"image/jpeg"
+	"image/png"
+	"strings"
+	"testing"
+
+	"github.com/easyai/easyai-ai-gateway/apps/api/internal/clients"
+	"github.com/easyai/easyai-ai-gateway/apps/api/internal/store"
+)
+
+func seedanceImageConstraintCandidate() store.RuntimeModelCandidate {
+	return store.RuntimeModelCandidate{
+		ModelType: "image_to_video",
+		Capabilities: map[string]any{
+			"image_to_video": map[string]any{
+				"input_image_resolution_range": map[string]any{
+					"min": map[string]any{"long_edge": float64(360), "short_edge": float64(360)},
+					"max": map[string]any{"long_edge": float64(1920), "short_edge": float64(1080)},
+				},
+				"input_image_aspect_ratio_range": []any{float64(0.39), float64(2.5)},
+			},
+		},
+	}
+}
+
+func pngImageDataURL(t *testing.T, width int, height int) string {
+	t.Helper()
+	source := image.NewRGBA(image.Rect(0, 0, width, height))
+	for y := 0; y < height; y++ {
+		for x := 0; x < width; x++ {
+			source.Set(x, y, color.RGBA{R: 80, G: 120, B: 160, A: 255})
+		}
+	}
+	var payload bytes.Buffer
+	if err := png.Encode(&payload, source); err != nil {
+		t.Fatal(err)
+	}
+	return "data:image/png;base64," + base64.StdEncoding.EncodeToString(payload.Bytes())
+}
+
+func decodedImageSize(t *testing.T, source string) image.Point {
+	t.Helper()
+	if !strings.HasPrefix(source, "data:image/jpeg;base64,") {
+		t.Fatalf("expected converted JPEG data URL, got prefix %.32q", source)
+	}
+	payload, err := base64.StdEncoding.DecodeString(strings.TrimPrefix(source, "data:image/jpeg;base64,"))
+	if err != nil {
+		t.Fatal(err)
+	}
+	config, err := jpeg.DecodeConfig(bytes.NewReader(payload))
+	if err != nil {
+		t.Fatal(err)
+	}
+	return image.Pt(config.Width, config.Height)
+}
+
+func TestNormalizeVideoInputImagesResizesOversizedSeedanceImage(t *testing.T) {
+	service := &Service{}
+	body := map[string]any{
+		"content": []any{
+			map[string]any{
+				"type":      "image_url",
+				"role":      "first_frame",
+				"image_url": map[string]any{"url": pngImageDataURL(t, 3840, 2160)},
+			},
+		},
+	}
+
+	normalized, err := service.normalizeVideoInputImages(context.Background(), body, seedanceImageConstraintCandidate())
+	if err != nil {
+		t.Fatal(err)
+	}
+	content := normalized["content"].([]any)
+	item := content[0].(map[string]any)
+	imageURL := item["image_url"].(map[string]any)["url"].(string)
+	if got := decodedImageSize(t, imageURL); got != image.Pt(1920, 1080) {
+		t.Fatalf("unexpected converted size: %v", got)
+	}
+}
+
+func TestNormalizeVideoInputImagesPadsAspectRatioViolation(t *testing.T) {
+	service := &Service{}
+	source := pngImageDataURL(t, 1530, 500)
+	body := map[string]any{"first_frame": source}
+
+	normalized, err := service.normalizeVideoInputImages(context.Background(), body, seedanceImageConstraintCandidate())
+	if err != nil {
+		t.Fatal(err)
+	}
+	if got := decodedImageSize(t, normalized["first_frame"].(string)); got != image.Pt(1530, 612) {
+		t.Fatalf("unexpected converted size: %v", got)
+	}
+}
+
+func TestNormalizeVideoInputImagesUpscalesBelowMinimumResolution(t *testing.T) {
+	service := &Service{}
+	body := map[string]any{"first_frame": pngImageDataURL(t, 200, 100)}
+
+	normalized, err := service.normalizeVideoInputImages(context.Background(), body, seedanceImageConstraintCandidate())
+	if err != nil {
+		t.Fatal(err)
+	}
+	if got := decodedImageSize(t, normalized["first_frame"].(string)); got != image.Pt(720, 360) {
+		t.Fatalf("unexpected converted size: %v", got)
+	}
+}
+
+func TestNormalizeVideoInputImagesKeepsCompliantImage(t *testing.T) {
+	service := &Service{}
+	source := pngImageDataURL(t, 1280, 720)
+	body := map[string]any{"first_frame": source}
+
+	normalized, err := service.normalizeVideoInputImages(context.Background(), body, seedanceImageConstraintCandidate())
+	if err != nil {
+		t.Fatal(err)
+	}
+	if normalized["first_frame"] != source {
+		t.Fatal("compliant image should remain unchanged")
+	}
+}
+
+func TestNormalizeVideoInputImagesReturnsActionableDecodeError(t *testing.T) {
+	service := &Service{}
+	body := map[string]any{"first_frame": "data:image/png;base64,bm90LWltYWdl"}
+
+	_, err := service.normalizeVideoInputImages(context.Background(), body, seedanceImageConstraintCandidate())
+	if err == nil {
+		t.Fatal("expected decode error")
+	}
+	if clients.ErrorCode(err) != "invalid_parameter" || clients.ErrorParam(err) != "first_frame" {
+		t.Fatalf("unexpected error contract: code=%s param=%s err=%v", clients.ErrorCode(err), clients.ErrorParam(err), err)
+	}
+}