adaptor.go 14 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480
  1. package jimeng
  2. import (
  3. "bytes"
  4. "crypto/hmac"
  5. "crypto/sha256"
  6. "encoding/base64"
  7. "encoding/hex"
  8. "fmt"
  9. "io"
  10. "net/http"
  11. "net/url"
  12. "sort"
  13. "strings"
  14. "time"
  15. "github.com/QuantumNous/new-api/common"
  16. "github.com/QuantumNous/new-api/model"
  17. "github.com/samber/lo"
  18. "github.com/gin-gonic/gin"
  19. "github.com/pkg/errors"
  20. "github.com/QuantumNous/new-api/constant"
  21. "github.com/QuantumNous/new-api/dto"
  22. "github.com/QuantumNous/new-api/relay/channel"
  23. taskcommon "github.com/QuantumNous/new-api/relay/channel/task/taskcommon"
  24. relaycommon "github.com/QuantumNous/new-api/relay/common"
  25. "github.com/QuantumNous/new-api/service"
  26. )
  27. // ============================
  28. // Request / Response structures
  29. // ============================
  30. type requestPayload struct {
  31. ReqKey string `json:"req_key"`
  32. BinaryDataBase64 []string `json:"binary_data_base64,omitempty"`
  33. ImageUrls []string `json:"image_urls,omitempty"`
  34. Prompt string `json:"prompt,omitempty"`
  35. Seed int64 `json:"seed"`
  36. AspectRatio string `json:"aspect_ratio"`
  37. Frames int `json:"frames,omitempty"`
  38. }
  39. type responsePayload struct {
  40. Code int `json:"code"`
  41. Message string `json:"message"`
  42. RequestId string `json:"request_id"`
  43. Data struct {
  44. TaskID string `json:"task_id"`
  45. } `json:"data"`
  46. }
  47. type responseTask struct {
  48. Code int `json:"code"`
  49. Data struct {
  50. BinaryDataBase64 []interface{} `json:"binary_data_base64"`
  51. ImageUrls interface{} `json:"image_urls"`
  52. RespData string `json:"resp_data"`
  53. Status string `json:"status"`
  54. VideoUrl string `json:"video_url"`
  55. } `json:"data"`
  56. Message string `json:"message"`
  57. RequestId string `json:"request_id"`
  58. Status int `json:"status"`
  59. TimeElapsed string `json:"time_elapsed"`
  60. }
  61. const (
  62. // 即梦限制单个文件最大4.7MB https://www.volcengine.com/docs/85621/1747301
  63. MaxFileSize int64 = 4*1024*1024 + 700*1024 // 4.7MB (4MB + 724KB)
  64. )
  65. // ============================
  66. // Adaptor implementation
  67. // ============================
  68. type TaskAdaptor struct {
  69. taskcommon.BaseBilling
  70. ChannelType int
  71. accessKey string
  72. secretKey string
  73. baseURL string
  74. }
  75. func (a *TaskAdaptor) Init(info *relaycommon.RelayInfo) {
  76. a.ChannelType = info.ChannelType
  77. a.baseURL = info.ChannelBaseUrl
  78. // apiKey format: "access_key|secret_key"
  79. keyParts := strings.Split(info.ApiKey, "|")
  80. if len(keyParts) == 2 {
  81. a.accessKey = strings.TrimSpace(keyParts[0])
  82. a.secretKey = strings.TrimSpace(keyParts[1])
  83. }
  84. }
  85. // ValidateRequestAndSetAction parses body, validates fields and sets default action.
  86. func (a *TaskAdaptor) ValidateRequestAndSetAction(c *gin.Context, info *relaycommon.RelayInfo) (taskErr *dto.TaskError) {
  87. return relaycommon.ValidateBasicTaskRequest(c, info, constant.TaskActionGenerate)
  88. }
  89. // BuildRequestURL constructs the upstream URL.
  90. func (a *TaskAdaptor) BuildRequestURL(info *relaycommon.RelayInfo) (string, error) {
  91. if isNewAPIRelay(info.ApiKey) {
  92. return fmt.Sprintf("%s/jimeng/?Action=CVSync2AsyncSubmitTask&Version=2022-08-31", a.baseURL), nil
  93. }
  94. return fmt.Sprintf("%s/?Action=CVSync2AsyncSubmitTask&Version=2022-08-31", a.baseURL), nil
  95. }
  96. // BuildRequestHeader sets required headers.
  97. func (a *TaskAdaptor) BuildRequestHeader(c *gin.Context, req *http.Request, info *relaycommon.RelayInfo) error {
  98. req.Header.Set("Content-Type", "application/json")
  99. req.Header.Set("Accept", "application/json")
  100. if isNewAPIRelay(info.ApiKey) {
  101. req.Header.Set("Authorization", "Bearer "+info.ApiKey)
  102. } else {
  103. return a.signRequest(req, a.accessKey, a.secretKey)
  104. }
  105. return nil
  106. }
  107. func (a *TaskAdaptor) BuildRequestBody(c *gin.Context, info *relaycommon.RelayInfo) (io.Reader, error) {
  108. v, exists := c.Get("task_request")
  109. if !exists {
  110. return nil, fmt.Errorf("request not found in context")
  111. }
  112. req, ok := v.(relaycommon.TaskSubmitReq)
  113. if !ok {
  114. return nil, fmt.Errorf("invalid request type in context")
  115. }
  116. // 支持openai sdk的图片上传方式
  117. if mf, err := c.MultipartForm(); err == nil {
  118. if files, exists := mf.File["input_reference"]; exists && len(files) > 0 {
  119. if len(files) == 1 {
  120. info.Action = constant.TaskActionGenerate
  121. } else if len(files) > 1 {
  122. info.Action = constant.TaskActionFirstTailGenerate
  123. }
  124. // 将上传的文件转换为base64格式
  125. var images []string
  126. for _, fileHeader := range files {
  127. // 检查文件大小
  128. if fileHeader.Size > MaxFileSize {
  129. return nil, fmt.Errorf("文件 %s 大小超过限制,最大允许 %d MB", fileHeader.Filename, MaxFileSize/(1024*1024))
  130. }
  131. file, err := fileHeader.Open()
  132. if err != nil {
  133. continue
  134. }
  135. fileBytes, err := io.ReadAll(file)
  136. file.Close()
  137. if err != nil {
  138. continue
  139. }
  140. // 将文件内容转换为base64
  141. base64Str := base64.StdEncoding.EncodeToString(fileBytes)
  142. images = append(images, base64Str)
  143. }
  144. req.Images = images
  145. }
  146. }
  147. body, err := a.convertToRequestPayload(&req)
  148. if err != nil {
  149. return nil, errors.Wrap(err, "convert request payload failed")
  150. }
  151. data, err := common.Marshal(body)
  152. if err != nil {
  153. return nil, err
  154. }
  155. return bytes.NewReader(data), nil
  156. }
  157. // DoRequest delegates to common helper.
  158. func (a *TaskAdaptor) DoRequest(c *gin.Context, info *relaycommon.RelayInfo, requestBody io.Reader) (*http.Response, error) {
  159. return channel.DoTaskApiRequest(a, c, info, requestBody)
  160. }
  161. // DoResponse handles upstream response, returns taskID etc.
  162. func (a *TaskAdaptor) DoResponse(c *gin.Context, resp *http.Response, info *relaycommon.RelayInfo) (taskID string, taskData []byte, taskErr *dto.TaskError) {
  163. responseBody, err := io.ReadAll(resp.Body)
  164. if err != nil {
  165. taskErr = service.TaskErrorWrapper(err, "read_response_body_failed", http.StatusInternalServerError)
  166. return
  167. }
  168. _ = resp.Body.Close()
  169. // Parse Jimeng response
  170. var jResp responsePayload
  171. if err := common.Unmarshal(responseBody, &jResp); err != nil {
  172. taskErr = service.TaskErrorWrapper(errors.Wrapf(err, "body: %s", responseBody), "unmarshal_response_body_failed", http.StatusInternalServerError)
  173. return
  174. }
  175. if jResp.Code != 10000 {
  176. taskErr = service.TaskErrorWrapper(fmt.Errorf("%s", jResp.Message), fmt.Sprintf("%d", jResp.Code), http.StatusInternalServerError)
  177. return
  178. }
  179. ov := dto.NewOpenAIVideo()
  180. ov.ID = info.PublicTaskID
  181. ov.TaskID = info.PublicTaskID
  182. ov.CreatedAt = time.Now().Unix()
  183. ov.Model = info.OriginModelName
  184. c.JSON(http.StatusOK, ov)
  185. return jResp.Data.TaskID, responseBody, nil
  186. }
  187. // FetchTask fetch task status
  188. func (a *TaskAdaptor) FetchTask(baseUrl, key string, body map[string]any, proxy string) (*http.Response, error) {
  189. taskID, ok := body["task_id"].(string)
  190. if !ok {
  191. return nil, fmt.Errorf("invalid task_id")
  192. }
  193. uri := fmt.Sprintf("%s/?Action=CVSync2AsyncGetResult&Version=2022-08-31", baseUrl)
  194. if isNewAPIRelay(key) {
  195. uri = fmt.Sprintf("%s/jimeng/?Action=CVSync2AsyncGetResult&Version=2022-08-31", a.baseURL)
  196. }
  197. payload := map[string]string{
  198. "req_key": "jimeng_vgfm_t2v_l20", // This is fixed value from doc: https://www.volcengine.com/docs/85621/1544774
  199. "task_id": taskID,
  200. }
  201. payloadBytes, err := common.Marshal(payload)
  202. if err != nil {
  203. return nil, errors.Wrap(err, "marshal fetch task payload failed")
  204. }
  205. req, err := http.NewRequest(http.MethodPost, uri, bytes.NewBuffer(payloadBytes))
  206. if err != nil {
  207. return nil, err
  208. }
  209. req.Header.Set("Accept", "application/json")
  210. req.Header.Set("Content-Type", "application/json")
  211. if isNewAPIRelay(key) {
  212. req.Header.Set("Authorization", "Bearer "+key)
  213. } else {
  214. keyParts := strings.Split(key, "|")
  215. if len(keyParts) != 2 {
  216. return nil, fmt.Errorf("invalid api key format for jimeng: expected 'ak|sk'")
  217. }
  218. accessKey := strings.TrimSpace(keyParts[0])
  219. secretKey := strings.TrimSpace(keyParts[1])
  220. if err := a.signRequest(req, accessKey, secretKey); err != nil {
  221. return nil, errors.Wrap(err, "sign request failed")
  222. }
  223. }
  224. client, err := service.GetHttpClientWithProxy(proxy)
  225. if err != nil {
  226. return nil, fmt.Errorf("new proxy http client failed: %w", err)
  227. }
  228. return client.Do(req)
  229. }
  230. func (a *TaskAdaptor) GetModelList() []string {
  231. return []string{"jimeng_vgfm_t2v_l20"}
  232. }
  233. func (a *TaskAdaptor) GetChannelName() string {
  234. return "jimeng"
  235. }
  236. func (a *TaskAdaptor) signRequest(req *http.Request, accessKey, secretKey string) error {
  237. var bodyBytes []byte
  238. var err error
  239. if req.Body != nil {
  240. bodyBytes, err = io.ReadAll(req.Body)
  241. if err != nil {
  242. return errors.Wrap(err, "read request body failed")
  243. }
  244. _ = req.Body.Close()
  245. req.Body = io.NopCloser(bytes.NewBuffer(bodyBytes)) // Rewind
  246. } else {
  247. bodyBytes = []byte{}
  248. }
  249. payloadHash := sha256.Sum256(bodyBytes)
  250. hexPayloadHash := hex.EncodeToString(payloadHash[:])
  251. t := time.Now().UTC()
  252. xDate := t.Format("20060102T150405Z")
  253. shortDate := t.Format("20060102")
  254. req.Header.Set("Host", req.URL.Host)
  255. req.Header.Set("X-Date", xDate)
  256. req.Header.Set("X-Content-Sha256", hexPayloadHash)
  257. // Sort and encode query parameters to create canonical query string
  258. queryParams := req.URL.Query()
  259. sortedKeys := make([]string, 0, len(queryParams))
  260. for k := range queryParams {
  261. sortedKeys = append(sortedKeys, k)
  262. }
  263. sort.Strings(sortedKeys)
  264. var queryParts []string
  265. for _, k := range sortedKeys {
  266. values := queryParams[k]
  267. sort.Strings(values)
  268. for _, v := range values {
  269. queryParts = append(queryParts, fmt.Sprintf("%s=%s", url.QueryEscape(k), url.QueryEscape(v)))
  270. }
  271. }
  272. canonicalQueryString := strings.Join(queryParts, "&")
  273. headersToSign := map[string]string{
  274. "host": req.URL.Host,
  275. "x-date": xDate,
  276. "x-content-sha256": hexPayloadHash,
  277. }
  278. if req.Header.Get("Content-Type") != "" {
  279. headersToSign["content-type"] = req.Header.Get("Content-Type")
  280. }
  281. var signedHeaderKeys []string
  282. for k := range headersToSign {
  283. signedHeaderKeys = append(signedHeaderKeys, k)
  284. }
  285. sort.Strings(signedHeaderKeys)
  286. var canonicalHeaders strings.Builder
  287. for _, k := range signedHeaderKeys {
  288. canonicalHeaders.WriteString(k)
  289. canonicalHeaders.WriteString(":")
  290. canonicalHeaders.WriteString(strings.TrimSpace(headersToSign[k]))
  291. canonicalHeaders.WriteString("\n")
  292. }
  293. signedHeaders := strings.Join(signedHeaderKeys, ";")
  294. canonicalRequest := fmt.Sprintf("%s\n%s\n%s\n%s\n%s\n%s",
  295. req.Method,
  296. req.URL.Path,
  297. canonicalQueryString,
  298. canonicalHeaders.String(),
  299. signedHeaders,
  300. hexPayloadHash,
  301. )
  302. hashedCanonicalRequest := sha256.Sum256([]byte(canonicalRequest))
  303. hexHashedCanonicalRequest := hex.EncodeToString(hashedCanonicalRequest[:])
  304. region := "cn-north-1"
  305. serviceName := "cv"
  306. credentialScope := fmt.Sprintf("%s/%s/%s/request", shortDate, region, serviceName)
  307. stringToSign := fmt.Sprintf("HMAC-SHA256\n%s\n%s\n%s",
  308. xDate,
  309. credentialScope,
  310. hexHashedCanonicalRequest,
  311. )
  312. kDate := hmacSHA256([]byte(secretKey), []byte(shortDate))
  313. kRegion := hmacSHA256(kDate, []byte(region))
  314. kService := hmacSHA256(kRegion, []byte(serviceName))
  315. kSigning := hmacSHA256(kService, []byte("request"))
  316. signature := hex.EncodeToString(hmacSHA256(kSigning, []byte(stringToSign)))
  317. authorization := fmt.Sprintf("HMAC-SHA256 Credential=%s/%s, SignedHeaders=%s, Signature=%s",
  318. accessKey,
  319. credentialScope,
  320. signedHeaders,
  321. signature,
  322. )
  323. req.Header.Set("Authorization", authorization)
  324. return nil
  325. }
  326. func hmacSHA256(key []byte, data []byte) []byte {
  327. h := hmac.New(sha256.New, key)
  328. h.Write(data)
  329. return h.Sum(nil)
  330. }
  331. func (a *TaskAdaptor) convertToRequestPayload(req *relaycommon.TaskSubmitReq) (*requestPayload, error) {
  332. r := requestPayload{
  333. ReqKey: req.Model,
  334. Prompt: req.Prompt,
  335. }
  336. switch req.Duration {
  337. case 10:
  338. r.Frames = 241 // 24*10+1 = 241
  339. default:
  340. r.Frames = 121 // 24*5+1 = 121
  341. }
  342. // Handle one-of image_urls or binary_data_base64
  343. if req.HasImage() {
  344. if strings.HasPrefix(req.Images[0], "http") {
  345. r.ImageUrls = req.Images
  346. } else {
  347. r.BinaryDataBase64 = req.Images
  348. }
  349. }
  350. if err := taskcommon.UnmarshalMetadata(req.Metadata, &r); err != nil {
  351. return nil, errors.Wrap(err, "unmarshal metadata failed")
  352. }
  353. // 即梦视频3.0 ReqKey转换
  354. // https://www.volcengine.com/docs/85621/1792707
  355. imageLen := lo.Max([]int{len(req.Images), len(r.BinaryDataBase64), len(r.ImageUrls)})
  356. if strings.Contains(r.ReqKey, "jimeng_v30") {
  357. if r.ReqKey == "jimeng_v30_pro" {
  358. // 3.0 pro只有固定的jimeng_ti2v_v30_pro
  359. r.ReqKey = "jimeng_ti2v_v30_pro"
  360. } else if imageLen > 1 {
  361. // 多张图片:首尾帧生成
  362. r.ReqKey = strings.TrimSuffix(strings.Replace(r.ReqKey, "jimeng_v30", "jimeng_i2v_first_tail_v30", 1), "p")
  363. } else if imageLen == 1 {
  364. // 单张图片:图生视频
  365. r.ReqKey = strings.TrimSuffix(strings.Replace(r.ReqKey, "jimeng_v30", "jimeng_i2v_first_v30", 1), "p")
  366. } else {
  367. // 无图片:文生视频
  368. r.ReqKey = strings.Replace(r.ReqKey, "jimeng_v30", "jimeng_t2v_v30", 1)
  369. }
  370. }
  371. return &r, nil
  372. }
  373. func (a *TaskAdaptor) ParseTaskResult(respBody []byte) (*relaycommon.TaskInfo, error) {
  374. resTask := responseTask{}
  375. if err := common.Unmarshal(respBody, &resTask); err != nil {
  376. return nil, errors.Wrap(err, "unmarshal task result failed")
  377. }
  378. taskResult := relaycommon.TaskInfo{}
  379. if resTask.Code == 10000 {
  380. taskResult.Code = 0
  381. } else {
  382. taskResult.Code = resTask.Code // todo uni code
  383. taskResult.Reason = resTask.Message
  384. taskResult.Status = model.TaskStatusFailure
  385. taskResult.Progress = "100%"
  386. }
  387. switch resTask.Data.Status {
  388. case "in_queue":
  389. taskResult.Status = model.TaskStatusQueued
  390. taskResult.Progress = "10%"
  391. case "done":
  392. taskResult.Status = model.TaskStatusSuccess
  393. taskResult.Progress = "100%"
  394. }
  395. taskResult.Url = resTask.Data.VideoUrl
  396. return &taskResult, nil
  397. }
  398. func (a *TaskAdaptor) ConvertToOpenAIVideo(originTask *model.Task) ([]byte, error) {
  399. var jimengResp responseTask
  400. if err := common.Unmarshal(originTask.Data, &jimengResp); err != nil {
  401. return nil, errors.Wrap(err, "unmarshal jimeng task data failed")
  402. }
  403. openAIVideo := dto.NewOpenAIVideo()
  404. openAIVideo.ID = originTask.TaskID
  405. openAIVideo.Status = originTask.Status.ToVideoStatus()
  406. openAIVideo.SetProgressStr(originTask.Progress)
  407. openAIVideo.SetMetadata("url", jimengResp.Data.VideoUrl)
  408. openAIVideo.CreatedAt = originTask.CreatedAt
  409. openAIVideo.CompletedAt = originTask.UpdatedAt
  410. if jimengResp.Code != 10000 {
  411. openAIVideo.Error = &dto.OpenAIVideoError{
  412. Message: jimengResp.Message,
  413. Code: fmt.Sprintf("%d", jimengResp.Code),
  414. }
  415. }
  416. return common.Marshal(openAIVideo)
  417. }
  418. func isNewAPIRelay(apiKey string) bool {
  419. return strings.HasPrefix(apiKey, "sk-")
  420. }