popiart-server
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 
 

624 lines
19 KiB

package server
import (
"bytes"
"context"
"encoding/base64"
"encoding/json"
"io"
"mime/multipart"
"net/http"
"net/http/httptest"
"os"
"path/filepath"
"strings"
"sync/atomic"
"testing"
"time"
)
func TestSubmitImageToVideoTaskUsesMultipartInputReference(t *testing.T) {
var (
gotModel string
gotPrompt string
gotSeconds string
gotSize string
gotBytes int
)
srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
if r.URL.Path != "/v1/videos" {
http.NotFound(w, r)
return
}
if err := r.ParseMultipartForm(4 << 20); err != nil {
t.Fatalf("parse multipart form: %v", err)
}
gotModel = r.FormValue("model")
gotPrompt = r.FormValue("prompt")
gotSeconds = r.FormValue("seconds")
gotSize = r.FormValue("size")
file, _, err := r.FormFile("input_reference")
if err != nil {
t.Fatalf("missing input_reference file: %v", err)
}
defer file.Close()
content, err := io.ReadAll(file)
if err != nil {
t.Fatalf("read uploaded file: %v", err)
}
gotBytes = len(content)
w.Header().Set("Content-Type", "application/json")
_, _ = w.Write([]byte(`{"id":"task_submit_123","status":"queued"}`))
}))
defer srv.Close()
client := newNewAPIClient(Config{NewAPIBaseURL: srv.URL})
taskID, err := client.submitImageToVideoTask(context.Background(), "sk-test", "sora-2", map[string]any{
"prompt": "animate the scene gently",
"duration_s": 5,
"aspect_ratio": "16:9",
}, imageEditReference{
Filename: "reference.png",
ContentType: "image/png",
Content: tinyPNG(t),
})
if err != nil {
t.Fatalf("submitImageToVideoTask: %v", err)
}
if taskID != "task_submit_123" {
t.Fatalf("expected task id task_submit_123, got %q", taskID)
}
if gotModel != "sora-2" {
t.Fatalf("expected model sora-2, got %q", gotModel)
}
if gotPrompt != "animate the scene gently" {
t.Fatalf("expected prompt to round-trip, got %q", gotPrompt)
}
if gotSeconds != "5" {
t.Fatalf("expected seconds 5, got %q", gotSeconds)
}
if gotSize != "1280x720" {
t.Fatalf("expected 16:9 to map to 1280x720, got %q", gotSize)
}
if gotBytes == 0 {
t.Fatal("expected uploaded reference bytes")
}
}
func TestSubmitImageToVideoTaskUsesURLImagesForViduModels(t *testing.T) {
var gotBody map[string]any
srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
if r.URL.Path != "/v1/videos" {
http.NotFound(w, r)
return
}
if got := r.Header.Get("Content-Type"); got != "application/json" {
t.Fatalf("expected application/json content type, got %q", got)
}
if err := json.NewDecoder(r.Body).Decode(&gotBody); err != nil {
t.Fatalf("decode request body: %v", err)
}
w.Header().Set("Content-Type", "application/json")
_, _ = w.Write([]byte(`{"id":"task_vidu_url_123","status":"queued"}`))
}))
defer srv.Close()
client := newNewAPIClient(Config{NewAPIBaseURL: srv.URL})
taskID, err := client.submitImageToVideoTask(context.Background(), "sk-test", "viduq2-pro-fast", map[string]any{
"prompt": "animate this still gently",
"duration_s": 5,
"aspect_ratio": "9:16",
}, imageEditReference{
Filename: "reference.png",
ContentType: "image/png",
URL: "https://media.popi.test/m/demo/reference.png",
})
if err != nil {
t.Fatalf("submitImageToVideoTask: %v", err)
}
if taskID != "task_vidu_url_123" {
t.Fatalf("expected task id task_vidu_url_123, got %q", taskID)
}
if gotBody["model"] != "viduq2-pro-fast" {
t.Fatalf("unexpected model: %#v", gotBody["model"])
}
images, ok := gotBody["images"].([]any)
if !ok || len(images) != 1 || images[0] != "https://media.popi.test/m/demo/reference.png" {
t.Fatalf("expected one image url, got %#v", gotBody["images"])
}
if gotBody["duration"] != float64(5) {
t.Fatalf("unexpected duration: %#v", gotBody["duration"])
}
metadata, ok := gotBody["metadata"].(map[string]any)
if !ok || metadata["aspect_ratio"] != "9:16" {
t.Fatalf("unexpected metadata: %#v", gotBody["metadata"])
}
}
func TestSubmitImageToVideoTaskUsesViduDefaultDurationWhenUnset(t *testing.T) {
var gotBody map[string]any
srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
if err := json.NewDecoder(r.Body).Decode(&gotBody); err != nil {
t.Fatalf("decode request body: %v", err)
}
w.Header().Set("Content-Type", "application/json")
_, _ = w.Write([]byte(`{"id":"task_vidu_default_duration","status":"queued"}`))
}))
defer srv.Close()
client := newNewAPIClient(Config{NewAPIBaseURL: srv.URL})
taskID, err := client.submitImageToVideoTask(context.Background(), "sk-test", "viduq3-turbo", map[string]any{
"prompt": "animate this still gently",
}, imageEditReference{
URL: "https://media.popi.test/m/demo/reference.png",
})
if err != nil {
t.Fatalf("submitImageToVideoTask: %v", err)
}
if taskID != "task_vidu_default_duration" {
t.Fatalf("unexpected task id: %q", taskID)
}
if gotBody["duration"] != float64(5) {
t.Fatalf("expected default vidu duration 5, got %#v", gotBody["duration"])
}
}
func TestSubmitMiniMaxVideoTaskUsesVideoGenerationsEndpoint(t *testing.T) {
var gotPath string
var gotBody map[string]any
srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
gotPath = r.URL.Path
if err := json.NewDecoder(r.Body).Decode(&gotBody); err != nil {
t.Fatalf("decode request body: %v", err)
}
w.Header().Set("Content-Type", "application/json")
_, _ = w.Write([]byte(`{"task_id":"task_minimax_video_123","status":"submitted"}`))
}))
defer srv.Close()
client := newNewAPIClient(Config{NewAPIBaseURL: srv.URL})
taskID, err := client.submitMiniMaxVideoTask(context.Background(), "sk-test", "I2V-01", map[string]any{
"prompt": "animate this portrait gently",
"size": "720x1280",
}, []imageEditReference{{
URL: "https://example.com/reference.png",
}})
if err != nil {
t.Fatalf("submitMiniMaxVideoTask: %v", err)
}
if taskID != "task_minimax_video_123" {
t.Fatalf("unexpected task id: %q", taskID)
}
if gotPath != "/v1/video/generations" {
t.Fatalf("expected /v1/video/generations, got %q", gotPath)
}
if gotBody["model"] != "I2V-01" {
t.Fatalf("unexpected model: %#v", gotBody["model"])
}
images, ok := gotBody["images"].([]any)
if !ok || len(images) != 1 || images[0] != "https://example.com/reference.png" {
t.Fatalf("unexpected images payload: %#v", gotBody["images"])
}
if gotBody["size"] != "720P" {
t.Fatalf("expected 720P resolution, got %#v", gotBody["size"])
}
}
func TestSubmitMiniMaxVideoTaskKeepsS2VImagesArray(t *testing.T) {
var gotBody map[string]any
srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
if err := json.NewDecoder(r.Body).Decode(&gotBody); err != nil {
t.Fatalf("decode request body: %v", err)
}
w.Header().Set("Content-Type", "application/json")
_, _ = w.Write([]byte(`{"task_id":"task_s2v_123","status":"submitted"}`))
}))
defer srv.Close()
client := newNewAPIClient(Config{NewAPIBaseURL: srv.URL})
taskID, err := client.submitMiniMaxVideoTask(context.Background(), "sk-test", "S2V-01", map[string]any{
"prompt": "animate the subject gently",
}, []imageEditReference{{
URL: "https://example.com/reference.png",
}})
if err != nil {
t.Fatalf("submitMiniMaxVideoTask: %v", err)
}
if taskID != "task_s2v_123" {
t.Fatalf("unexpected task id: %q", taskID)
}
images, ok := gotBody["images"].([]any)
if !ok || len(images) != 1 || images[0] != "https://example.com/reference.png" {
t.Fatalf("unexpected images payload for S2V: %#v", gotBody["images"])
}
}
func TestResolveVideoReferencesSupportsImagesArray(t *testing.T) {
cfg := Config{
SQLitePath: filepath.Join(t.TempDir(), "popiart.db"),
SkillhubDir: makeEmptySkillhub(t),
SessionSecret: "test-secret",
}
server, err := NewWithConfig(cfg)
if err != nil {
t.Fatalf("NewWithConfig: %v", err)
}
imageBytes := tinyPNG(t)
refSrv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
w.Header().Set("Content-Type", "image/png")
_, _ = w.Write(imageBytes)
}))
defer refSrv.Close()
refs, err := server.resolveVideoReferences(context.Background(), &job{
UserID: "user_test",
UpstreamKey: "sk-upstream",
}, map[string]any{
"images": []any{refSrv.URL + "/a.png", refSrv.URL + "/b.png"},
})
if err != nil {
t.Fatalf("resolveVideoReferences: %v", err)
}
if len(refs) != 2 {
t.Fatalf("expected 2 refs, got %#v", refs)
}
}
func TestResolveVideoReferencesAcceptsCanonicalImageURL(t *testing.T) {
cfg := Config{
SQLitePath: filepath.Join(t.TempDir(), "popiart.db"),
SkillhubDir: makeEmptySkillhub(t),
SessionSecret: "test-secret",
}
server, err := NewWithConfig(cfg)
if err != nil {
t.Fatalf("NewWithConfig: %v", err)
}
imageBytes := tinyPNG(t)
refSrv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
w.Header().Set("Content-Type", "image/png")
_, _ = w.Write(imageBytes)
}))
defer refSrv.Close()
refs, err := server.resolveVideoReferences(context.Background(), &job{
UserID: "user_test",
UpstreamKey: "sk-upstream",
}, map[string]any{
"image_url": refSrv.URL + "/single.png",
})
if err != nil {
t.Fatalf("resolveVideoReferences canonical image_url: %v", err)
}
if len(refs) != 1 {
t.Fatalf("expected 1 ref, got %#v", refs)
}
if refs[0].URL != refSrv.URL+"/single.png" {
t.Fatalf("expected URL to round-trip, got %q", refs[0].URL)
}
if refs[0].ContentType != "image/png" {
t.Fatalf("expected image/png, got %q", refs[0].ContentType)
}
if !bytes.Equal(refs[0].Content, imageBytes) {
t.Fatal("expected canonical image_url bytes to match")
}
}
func TestResolveVideoReferencesAcceptsSignedSameOriginMediaURL(t *testing.T) {
cfg := Config{
SQLitePath: filepath.Join(t.TempDir(), "popiart.db"),
SkillhubDir: makeEmptySkillhub(t),
SessionSecret: "test-secret",
}
server, err := NewWithConfig(cfg)
if err != nil {
t.Fatalf("NewWithConfig: %v", err)
}
srv := httptest.NewServer(server.Handler())
defer srv.Close()
server.cfg.PublicBaseURL = srv.URL
sessionToken, _, ok, err := server.store.createSession("sk-video-same-origin-user")
if err != nil {
t.Fatalf("createSession: %v", err)
}
if !ok {
t.Fatal("expected session creation to succeed")
}
current, exists, err := server.store.session(sessionToken)
if err != nil {
t.Fatalf("load session: %v", err)
}
if !exists {
t.Fatal("expected stored session")
}
imageBytes := tinyPNG(t)
var body bytes.Buffer
writer := multipart.NewWriter(&body)
part, err := writer.CreateFormFile("file", "video-reference.png")
if err != nil {
t.Fatalf("create form file: %v", err)
}
if _, err := part.Write(imageBytes); err != nil {
t.Fatalf("write form file: %v", err)
}
if err := writer.Close(); err != nil {
t.Fatalf("close multipart writer: %v", err)
}
req := httptest.NewRequest(http.MethodPost, "/v1/media/upload", &body)
req.Header.Set("Authorization", "Bearer "+sessionToken)
req.Header.Set("Content-Type", writer.FormDataContentType())
rec := httptest.NewRecorder()
server.Handler().ServeHTTP(rec, req)
if rec.Code != http.StatusCreated {
t.Fatalf("expected media upload 201, got %d body=%s", rec.Code, rec.Body.String())
}
var envelope struct {
OK bool `json:"ok"`
Data media `json:"data"`
}
if err := json.Unmarshal(rec.Body.Bytes(), &envelope); err != nil {
t.Fatalf("decode media upload response: %v body=%s", err, rec.Body.String())
}
if !envelope.OK || envelope.Data.URL == "" {
t.Fatalf("expected signed media url, got %s", rec.Body.String())
}
refs, err := server.resolveVideoReferences(context.Background(), &job{
UserID: current.User.ID,
SessionID: current.Token,
UpstreamKey: current.UpstreamKey,
}, map[string]any{
"image_url": envelope.Data.URL,
})
if err != nil {
t.Fatalf("resolveVideoReferences same-origin signed media url: %v", err)
}
if len(refs) != 1 {
t.Fatalf("expected 1 ref, got %#v", refs)
}
if refs[0].URL != envelope.Data.URL {
t.Fatalf("expected signed URL to round-trip, got %q", refs[0].URL)
}
if refs[0].ContentType != "image/png" {
t.Fatalf("expected image/png, got %q", refs[0].ContentType)
}
if !bytes.Equal(refs[0].Content, imageBytes) {
t.Fatal("expected same-origin signed media bytes to match")
}
}
func TestFetchVideoTaskFallsBackToGenericTaskEnvelope(t *testing.T) {
srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
switch r.URL.Path {
case "/v1/videos/task_fallback_123":
http.NotFound(w, r)
case "/v1/video/generations/task_fallback_123":
w.Header().Set("Content-Type", "application/json")
_, _ = w.Write([]byte(`{"code":"success","message":"","data":{"task_id":"task_fallback_123","status":"SUCCESS","result_url":"http://example.com/out.mp4","progress":"100%"}}`))
default:
http.NotFound(w, r)
}
}))
defer srv.Close()
client := newNewAPIClient(Config{NewAPIBaseURL: srv.URL})
result, err := client.fetchVideoTask(context.Background(), "sk-test", "task_fallback_123")
if err != nil {
t.Fatalf("fetchVideoTask: %v", err)
}
if result.TaskID != "task_fallback_123" {
t.Fatalf("expected task id task_fallback_123, got %q", result.TaskID)
}
if result.Status != "completed" {
t.Fatalf("expected normalized completed status, got %q", result.Status)
}
if result.URL != "http://example.com/out.mp4" {
t.Fatalf("expected result url to round-trip, got %q", result.URL)
}
}
func TestExecuteImageToVideoJobCompletesAndArtifactCanBeRead(t *testing.T) {
refBytes := tinyPNG(t)
videoBytes := []byte("not-a-real-mp4-but-good-enough-for-streaming")
var (
fetchCalls int32
contentAuth string
)
var localProxyURL string
srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
switch r.URL.Path {
case "/input.png":
w.Header().Set("Content-Type", "image/png")
_, _ = w.Write(refBytes)
case "/v1/videos":
if err := r.ParseMultipartForm(4 << 20); err != nil {
t.Fatalf("parse multipart form: %v", err)
}
if got := r.FormValue("model"); got != "sora-2" {
t.Fatalf("expected model sora-2, got %q", got)
}
file, _, err := r.FormFile("input_reference")
if err != nil {
t.Fatalf("missing input_reference: %v", err)
}
defer file.Close()
if _, err := io.ReadAll(file); err != nil {
t.Fatalf("read input_reference: %v", err)
}
w.Header().Set("Content-Type", "application/json")
_, _ = w.Write([]byte(`{"id":"task_video_123","status":"queued"}`))
case "/v1/videos/task_video_123":
call := atomic.AddInt32(&fetchCalls, 1)
w.Header().Set("Content-Type", "application/json")
if call == 1 {
_, _ = w.Write([]byte(`{"id":"task_video_123","status":"in_progress","progress":30}`))
return
}
_, _ = w.Write([]byte(`{"id":"task_video_123","status":"completed","progress":100,"metadata":{"url":"` + localProxyURL + `","format":"mp4"}}`))
case "/v1/videos/task_video_123/content":
contentAuth = r.Header.Get("Authorization")
w.Header().Set("Content-Type", "video/mp4")
_, _ = w.Write(videoBytes)
default:
http.NotFound(w, r)
}
}))
defer srv.Close()
localProxyURL = strings.Replace(srv.URL, "127.0.0.1", "localhost", 1) + "/v1/videos/task_video_123/content"
cfg := Config{
NewAPIBaseURL: srv.URL,
SQLitePath: filepath.Join(t.TempDir(), "popiart.db"),
SkillhubDir: makeEmptySkillhub(t),
SessionSecret: "test-secret",
}
server, err := NewWithConfig(cfg)
if err != nil {
t.Fatalf("NewWithConfig: %v", err)
}
server.cfg.PublicBaseURL = "http://127.0.0.1:8080"
token, _, ok, err := server.store.createSession("sk-test-upstream")
if err != nil {
t.Fatalf("createSession: %v", err)
}
if !ok {
t.Fatal("expected session creation to succeed")
}
current, exists, err := server.store.session(token)
if err != nil {
t.Fatalf("load session: %v", err)
}
if !exists {
t.Fatal("expected stored session")
}
record, _, err := server.store.createJob(
"popiskill-video-image2video-basic-v1",
"video.image2video",
"sora-2",
routeExecMode("video.image2video"),
map[string]any{
"image_url": srv.URL + "/input.png",
"motion_prompt": "the camera slowly pushes in",
"duration_s": 4,
"aspect_ratio": "9:16",
},
"",
"normal",
"",
current,
)
if err != nil {
t.Fatalf("createJob: %v", err)
}
server.executeImageToVideoJob(record)
done, exists, err := server.store.getJob(current.User.ID, record.JobID)
if err != nil {
t.Fatalf("getJob: %v", err)
}
if !exists {
t.Fatal("expected completed job")
}
if done.Status != "done" {
t.Fatalf("expected job status done, got %q", done.Status)
}
if done.NewAPITaskID != "task_video_123" {
t.Fatalf("expected persisted task id task_video_123, got %q", done.NewAPITaskID)
}
if len(done.ArtifactIDs) != 1 {
t.Fatalf("expected one artifact, got %d", len(done.ArtifactIDs))
}
item, ref, exists, err := server.store.artifactRef(current.User.ID, done.ArtifactIDs[0])
if err != nil {
t.Fatalf("artifactRef: %v", err)
}
if !exists {
t.Fatal("expected artifact ref")
}
if item.ContentType != "video/mp4" {
t.Fatalf("expected video/mp4 artifact, got %q", item.ContentType)
}
if item.MediaID == "" {
t.Fatalf("expected media id on persisted artifact, got %#v", item)
}
if item.URL == "" {
t.Fatalf("expected stable url on persisted artifact, got %#v", item)
}
if item.StorageStatus != "ready" {
t.Fatalf("expected ready storage status, got %#v", item.StorageStatus)
}
if !strings.HasSuffix(item.Filename, ".mp4") {
t.Fatalf("expected .mp4 filename, got %q", item.Filename)
}
ctx, cancel := context.WithTimeout(context.Background(), time.Second)
defer cancel()
contentType, _, reader, err := server.newapi.openResultRef(ctx, current.UpstreamKey, ref)
if err != nil {
t.Fatalf("openResultRef: %v", err)
}
defer reader.Close()
content, err := io.ReadAll(reader)
if err != nil {
t.Fatalf("read video content: %v", err)
}
if contentType != "video/mp4" {
t.Fatalf("expected fetched content type video/mp4, got %q", contentType)
}
if string(content) != string(videoBytes) {
t.Fatalf("unexpected streamed content: %q", string(content))
}
if contentAuth != "Bearer sk-test-upstream" {
t.Fatalf("expected auth header to be forwarded to proxy content URL, got %q", contentAuth)
}
}
func TestResolveVideoSizeUsesViduResolutionLabels(t *testing.T) {
if got := resolveVideoSize("viduq2", map[string]any{"aspect_ratio": "16:9"}, imageEditReference{}); got != "720p" {
t.Fatalf("expected vidu 16:9 to map to 720p, got %q", got)
}
if got := resolveVideoSize("viduq2", map[string]any{"size": "1080p"}, imageEditReference{}); got != "1080p" {
t.Fatalf("expected explicit vidu resolution to be preserved, got %q", got)
}
if got := resolveVideoSize("sora-2", map[string]any{"aspect_ratio": "16:9"}, imageEditReference{}); got != "1280x720" {
t.Fatalf("expected sora size to remain pixel dimensions, got %q", got)
}
}
func tinyPNG(t *testing.T) []byte {
t.Helper()
data, err := base64.StdEncoding.DecodeString("iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAQAAAC1HAwCAAAAC0lEQVR42mP8/x8AAwMCAO+aH8kAAAAASUVORK5CYII=")
if err != nil {
t.Fatalf("decode tiny png: %v", err)
}
return data
}
func makeEmptySkillhub(t *testing.T) string {
t.Helper()
dir := t.TempDir()
if err := os.WriteFile(filepath.Join(dir, "index.json"), []byte(`{"version":1,"skills":[]}`), 0o644); err != nil {
t.Fatalf("write skillhub index: %v", err)
}
return dir
}