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.

289 lines
8.3 KiB

6 years ago
3 years ago
5 years ago
5 years ago
3 years ago
3 years ago
3 years ago
5 years ago
3 years ago
5 years ago
3 years ago
3 years ago
4 years ago
3 years ago
4 years ago
3 years ago
3 years ago
3 years ago
3 years ago
3 years ago
3 years ago
3 years ago
3 years ago
5 years ago
3 years ago
3 years ago
3 years ago
5 years ago
3 years ago
12 years ago
  1. package operation
  2. import (
  3. "bytes"
  4. "encoding/json"
  5. "fmt"
  6. "io"
  7. "io/ioutil"
  8. "mime"
  9. "mime/multipart"
  10. "net/http"
  11. "net/textproto"
  12. "path/filepath"
  13. "strings"
  14. "time"
  15. "github.com/chrislusf/seaweedfs/weed/glog"
  16. "github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
  17. "github.com/chrislusf/seaweedfs/weed/security"
  18. "github.com/chrislusf/seaweedfs/weed/util"
  19. )
  20. type UploadOption struct {
  21. UploadUrl string
  22. Filename string
  23. Cipher bool
  24. IsInputCompressed bool
  25. MimeType string
  26. PairMap map[string]string
  27. Jwt security.EncodedJwt
  28. }
  29. type UploadResult struct {
  30. Name string `json:"name,omitempty"`
  31. Size uint32 `json:"size,omitempty"`
  32. Error string `json:"error,omitempty"`
  33. ETag string `json:"eTag,omitempty"`
  34. CipherKey []byte `json:"cipherKey,omitempty"`
  35. Mime string `json:"mime,omitempty"`
  36. Gzip uint32 `json:"gzip,omitempty"`
  37. ContentMd5 string `json:"contentMd5,omitempty"`
  38. RetryCount int `json:"-"`
  39. }
  40. func (uploadResult *UploadResult) ToPbFileChunk(fileId string, offset int64) *filer_pb.FileChunk {
  41. fid, _ := filer_pb.ToFileIdObject(fileId)
  42. return &filer_pb.FileChunk{
  43. FileId: fileId,
  44. Offset: offset,
  45. Size: uint64(uploadResult.Size),
  46. Mtime: time.Now().UnixNano(),
  47. ETag: uploadResult.ContentMd5,
  48. CipherKey: uploadResult.CipherKey,
  49. IsCompressed: uploadResult.Gzip > 0,
  50. Fid: fid,
  51. }
  52. }
  53. // HTTPClient interface for testing
  54. type HTTPClient interface {
  55. Do(req *http.Request) (*http.Response, error)
  56. }
  57. var (
  58. HttpClient HTTPClient
  59. )
  60. func init() {
  61. HttpClient = &http.Client{Transport: &http.Transport{
  62. MaxIdleConns: 1024,
  63. MaxIdleConnsPerHost: 1024,
  64. }}
  65. }
  66. var fileNameEscaper = strings.NewReplacer(`\`, `\\`, `"`, `\"`)
  67. // Upload sends a POST request to a volume server to upload the content with adjustable compression level
  68. func UploadData(data []byte, option *UploadOption) (uploadResult *UploadResult, err error) {
  69. uploadResult, err = retriedUploadData(data, option)
  70. return
  71. }
  72. // Upload sends a POST request to a volume server to upload the content with fast compression
  73. func Upload(reader io.Reader, option *UploadOption) (uploadResult *UploadResult, err error, data []byte) {
  74. uploadResult, err, data = doUpload(reader, option)
  75. return
  76. }
  77. func doUpload(reader io.Reader, option *UploadOption) (uploadResult *UploadResult, err error, data []byte) {
  78. bytesReader, ok := reader.(*util.BytesReader)
  79. if ok {
  80. data = bytesReader.Bytes
  81. } else {
  82. data, err = ioutil.ReadAll(reader)
  83. if err != nil {
  84. err = fmt.Errorf("read input: %v", err)
  85. return
  86. }
  87. }
  88. uploadResult, uploadErr := retriedUploadData(data, option)
  89. return uploadResult, uploadErr, data
  90. }
  91. func retriedUploadData(data []byte, option *UploadOption) (uploadResult *UploadResult, err error) {
  92. for i := 0; i < 3; i++ {
  93. uploadResult, err = doUploadData(data, option)
  94. if err == nil {
  95. uploadResult.RetryCount = i
  96. return
  97. } else {
  98. glog.Warningf("uploading to %s: %v", option.UploadUrl, err)
  99. }
  100. time.Sleep(time.Millisecond * time.Duration(237*(i+1)))
  101. }
  102. return
  103. }
  104. func doUploadData(data []byte, option *UploadOption) (uploadResult *UploadResult, err error) {
  105. contentIsGzipped := option.IsInputCompressed
  106. shouldGzipNow := false
  107. if !option.IsInputCompressed {
  108. if option.MimeType == "" {
  109. option.MimeType = http.DetectContentType(data)
  110. // println("detect1 mimetype to", MimeType)
  111. if option.MimeType == "application/octet-stream" {
  112. option.MimeType = ""
  113. }
  114. }
  115. if shouldBeCompressed, iAmSure := util.IsCompressableFileType(filepath.Base(option.Filename), option.MimeType); iAmSure && shouldBeCompressed {
  116. shouldGzipNow = true
  117. } else if !iAmSure && option.MimeType == "" && len(data) > 16*1024 {
  118. var compressed []byte
  119. compressed, err = util.GzipData(data[0:128])
  120. shouldGzipNow = len(compressed)*10 < 128*9 // can not compress to less than 90%
  121. }
  122. }
  123. var clearDataLen int
  124. // gzip if possible
  125. // this could be double copying
  126. clearDataLen = len(data)
  127. clearData := data
  128. if shouldGzipNow && !option.Cipher {
  129. compressed, compressErr := util.GzipData(data)
  130. // fmt.Printf("data is compressed from %d ==> %d\n", len(data), len(compressed))
  131. if compressErr == nil {
  132. data = compressed
  133. contentIsGzipped = true
  134. }
  135. } else if option.IsInputCompressed {
  136. // just to get the clear data length
  137. clearData, err = util.DecompressData(data)
  138. if err == nil {
  139. clearDataLen = len(clearData)
  140. }
  141. }
  142. if option.Cipher {
  143. // encrypt(gzip(data))
  144. // encrypt
  145. cipherKey := util.GenCipherKey()
  146. encryptedData, encryptionErr := util.Encrypt(clearData, cipherKey)
  147. if encryptionErr != nil {
  148. err = fmt.Errorf("encrypt input: %v", encryptionErr)
  149. return
  150. }
  151. // upload data
  152. uploadResult, err = upload_content(option.UploadUrl, func(w io.Writer) (err error) {
  153. _, err = w.Write(encryptedData)
  154. return
  155. }, "", false, len(encryptedData), "", nil, option.Jwt)
  156. if uploadResult == nil {
  157. return
  158. }
  159. uploadResult.Name = option.Filename
  160. uploadResult.Mime = option.MimeType
  161. uploadResult.CipherKey = cipherKey
  162. uploadResult.Size = uint32(clearDataLen)
  163. } else {
  164. // upload data
  165. uploadResult, err = upload_content(option.UploadUrl, func(w io.Writer) (err error) {
  166. _, err = w.Write(data)
  167. return
  168. }, option.Filename, contentIsGzipped, len(data), option.MimeType, option.PairMap, option.Jwt)
  169. if uploadResult == nil {
  170. return
  171. }
  172. uploadResult.Size = uint32(clearDataLen)
  173. if contentIsGzipped {
  174. uploadResult.Gzip = 1
  175. }
  176. }
  177. return uploadResult, err
  178. }
  179. func upload_content(uploadUrl string, fillBufferFunction func(w io.Writer) error, filename string, isGzipped bool, originalDataSize int, mtype string, pairMap map[string]string, jwt security.EncodedJwt) (*UploadResult, error) {
  180. buf := GetBuffer()
  181. defer PutBuffer(buf)
  182. body_writer := multipart.NewWriter(buf)
  183. h := make(textproto.MIMEHeader)
  184. h.Set("Content-Disposition", fmt.Sprintf(`form-data; name="file"; filename="%s"`, fileNameEscaper.Replace(filename)))
  185. h.Set("Idempotency-Key", uploadUrl)
  186. if mtype == "" {
  187. mtype = mime.TypeByExtension(strings.ToLower(filepath.Ext(filename)))
  188. }
  189. if mtype != "" {
  190. h.Set("Content-Type", mtype)
  191. }
  192. if isGzipped {
  193. h.Set("Content-Encoding", "gzip")
  194. }
  195. file_writer, cp_err := body_writer.CreatePart(h)
  196. if cp_err != nil {
  197. glog.V(0).Infoln("error creating form file", cp_err.Error())
  198. return nil, cp_err
  199. }
  200. if err := fillBufferFunction(file_writer); err != nil {
  201. glog.V(0).Infoln("error copying data", err)
  202. return nil, err
  203. }
  204. content_type := body_writer.FormDataContentType()
  205. if err := body_writer.Close(); err != nil {
  206. glog.V(0).Infoln("error closing body", err)
  207. return nil, err
  208. }
  209. req, postErr := http.NewRequest("POST", uploadUrl, bytes.NewReader(buf.Bytes()))
  210. if postErr != nil {
  211. glog.V(1).Infof("create upload request %s: %v", uploadUrl, postErr)
  212. return nil, fmt.Errorf("create upload request %s: %v", uploadUrl, postErr)
  213. }
  214. req.Header.Set("Content-Type", content_type)
  215. for k, v := range pairMap {
  216. req.Header.Set(k, v)
  217. }
  218. if jwt != "" {
  219. req.Header.Set("Authorization", "BEARER "+string(jwt))
  220. }
  221. // print("+")
  222. resp, post_err := HttpClient.Do(req)
  223. if post_err != nil {
  224. if strings.Contains(post_err.Error(), "connection reset by peer") ||
  225. strings.Contains(post_err.Error(), "use of closed network connection") {
  226. resp, post_err = HttpClient.Do(req)
  227. }
  228. }
  229. if post_err != nil {
  230. return nil, fmt.Errorf("upload %s %d bytes to %v: %v", filename, originalDataSize, uploadUrl, post_err)
  231. }
  232. // print("-")
  233. defer util.CloseResponse(resp)
  234. var ret UploadResult
  235. etag := getEtag(resp)
  236. if resp.StatusCode == http.StatusNoContent {
  237. ret.ETag = etag
  238. return &ret, nil
  239. }
  240. resp_body, ra_err := ioutil.ReadAll(resp.Body)
  241. if ra_err != nil {
  242. return nil, fmt.Errorf("read response body %v: %v", uploadUrl, ra_err)
  243. }
  244. unmarshal_err := json.Unmarshal(resp_body, &ret)
  245. if unmarshal_err != nil {
  246. glog.Errorf("unmarshal %s: %v", uploadUrl, string(resp_body))
  247. return nil, fmt.Errorf("unmarshal %v: %v", uploadUrl, unmarshal_err)
  248. }
  249. if ret.Error != "" {
  250. return nil, fmt.Errorf("unmarshalled error %v: %v", uploadUrl, ret.Error)
  251. }
  252. ret.ETag = etag
  253. ret.ContentMd5 = resp.Header.Get("Content-MD5")
  254. return &ret, nil
  255. }
  256. func getEtag(r *http.Response) (etag string) {
  257. etag = r.Header.Get("ETag")
  258. if strings.HasPrefix(etag, "\"") && strings.HasSuffix(etag, "\"") {
  259. etag = etag[1 : len(etag)-1]
  260. }
  261. return
  262. }