func writeToOss(h HadoopFile, ossDir string, hdfsClient *hdfs.Client,
ossClient *oss.Client, bucket string, b bool) error {
log.Printf("Start to sync %s to ossDir %s\n", h.dirPrefix+h.fileInfo.Name(), ossDir)
hdfsPath := h.dirPrefix + h.fileInfo.Name()
timeStamp := fmt.Sprintf("%d", makeTimestamp())
ossPath := ossDir + timeStamp + h.fileInfo.Name()
if b {
dateStr, err := getDateStamp(hdfsPath)
if err != nil {
log.Printf("get date info from hdfsPath %s failed!", hdfsPath)
return err
}
ossPath = dateStr + "/" + timeStamp + h.fileInfo.Name()
if err != nil {
return err
}
}
reader, err := hdfsClient.Open(hdfsPath)
if err != nil {
log.Printf("hdfsClient opend failed! err message: %s\n", err)
return err
}
defer reader.Close()
err = ossClient.PutObject(bucket, ossPath, reader, nil)
if err != nil {
log.Printf("Oss Append Object failed %s!\n", err)
return err
}
log.Printf("Finished sync %s to ossDir \n", h.dirPrefix+h.fileInfo.Name(), ossDir)
return nil
}
func walkDir(client *hdfs.Client, dir string, visit walkFunc) error {
dirReader, err := client.Open(dir)
if err != nil {
return err
}
var partial []os.FileInfo
for ; err != io.EOF; partial, err = dirReader.Readdir(100) {
if err != nil {
return err
}
for _, child := range partial {
childPath := path.Join(dir, child.Name())
visit(childPath, child)
if child.IsDir() {
err = walkDir(client, childPath, visit)
if err != nil {
return err
}
}
}
}
return nil
}
func findStacktrace(client *hdfs.Client, name string) (string, error) {
log.Println("Reading", name)
file, err := client.Open(name)
if err != nil {
return "", err
}
data, err := ioutil.ReadAll(file)
if err != nil {
return "", err
}
var logs [][]byte
lines := bytes.SplitAfter(data, []byte("\n"))
for _, line := range lines {
matched := false
for _, token := range logsToSkip {
if bytes.Contains(line, token) {
matched = true
break
}
}
if !matched {
logs = append(logs, line)
}
}
log.Println("Finished", name)
return string(bytes.Join(logs, nil)), nil
}
// expandGlobs recursively expands globs in a filepath. It assumes the paths
// are already cleaned and normalize (ie, absolute).
func expandGlobs(client *hdfs.Client, globbedPath string) ([]string, error) {
parts := strings.Split(globbedPath, "/")[1:]
var res []string
var splitAt int
for splitAt = range parts {
if hasGlob(parts[splitAt]) {
break
}
}
var base, glob, next, remainder string
base = "/" + path.Join(parts[:splitAt]...)
glob = parts[splitAt]
if len(parts) > splitAt+1 {
next = parts[splitAt+1]
remainder = path.Join(parts[splitAt+2:]...)
} else {
next = ""
remainder = ""
}
list, err := client.ReadDir(base)
if err != nil {
return nil, err
}
for _, fi := range list {
match, _ := path.Match(glob, fi.Name())
if !match {
continue
}
if !hasGlob(next) {
_, err := client.Stat(path.Join(base, fi.Name(), next))
if err != nil && !os.IsNotExist(err) {
return nil, err
} else if os.IsNotExist(err) {
continue
}
}
newPath := path.Join(base, fi.Name(), next, remainder)
if hasGlob(newPath) {
children, err := expandGlobs(client, newPath)
if err != nil {
return nil, err
}
res = append(res, children...)
} else {
res = append(res, newPath)
}
}
return res, nil
}
func HadoopReadWriteGzipFile(f F, h HadoopFile, hdfsClient *hdfs.Client) (int64, error) {
hdfsPath := h.dirPrefix + h.fileInfo.Name()
hdfsReader, err := hdfsClient.Open(hdfsPath)
if err != nil {
log.Printf("hdfsClient opend failed! err message: %s\n", err)
return 0, err
}
defer hdfsReader.Close()
log.Printf("start to use io Copy \n")
written, err := io.Copy(f.fw, hdfsReader)
return written, err
}
func moveTo(client *hdfs.Client, source, dest string, force bool) {
if force {
err := client.Remove(dest)
if err != nil && !os.IsNotExist(err) {
fatal(err)
}
}
err := client.Rename(source, dest)
if err != nil {
fatal(err)
}
}
func walk(client *hdfs.Client, root string, visit walkFunc) error {
rootInfo, err := client.Stat(root)
if err != nil {
return err
}
visit(root, rootInfo)
if rootInfo.IsDir() {
err = walkDir(client, root, visit)
if err != nil {
return err
}
}
return nil
}
func RecurseInfos(hdfsClient *hdfs.Client, hdfsDir string) ([]HadoopFile, error) {
hdfsFiles := make([]HadoopFile, 0)
dirs, err := hdfsClient.ReadDir(hdfsDir)
if err != nil {
return nil, err
}
for _, fileInfo := range dirs {
if fileInfo.IsDir() {
recurseInfos, err := RecurseInfos(hdfsClient, hdfsDir+fileInfo.Name()+"/")
if err != nil {
return nil, err
}
for _, f := range recurseInfos {
hdfsFiles = append(hdfsFiles, f)
}
} else {
f := HadoopFile{dirPrefix: hdfsDir, fileInfo: fileInfo}
hdfsFiles = append(hdfsFiles, f)
}
}
return hdfsFiles, nil
}
func duDir(client *hdfs.Client, tw *tabwriter.Writer, dir string, humanReadable bool) int64 {
dirReader, err := client.Open(dir)
if err != nil {
fmt.Fprintln(os.Stderr, err)
return 0
}
var partial []os.FileInfo
var dirSize int64
for ; err != io.EOF; partial, err = dirReader.Readdir(100) {
if err != nil {
fmt.Fprintln(os.Stderr, err)
return dirSize
}
for _, child := range partial {
childPath := path.Join(dir, child.Name())
info, err := client.Stat(childPath)
if err != nil {
fmt.Fprintln(os.Stderr, err)
return 0
}
var size int64
if info.IsDir() {
size = duDir(client, tw, childPath, humanReadable)
} else {
size = info.Size()
}
printSize(tw, size, childPath, humanReadable)
dirSize += size
}
}
return dirSize
}
func moveTo(client *hdfs.Client, source, dest string, force bool) {
resp, err := client.Stat(dest)
if force {
if err == nil && resp.IsDir() {
if err = client.Remove(dest); err != nil {
fatal(err)
}
}
} else if err == nil {
fatal(&os.PathError{"rename", dest, os.ErrExist})
}
err = client.Rename(source, dest)
if err != nil {
fatal(err)
}
}
func printDir(client *hdfs.Client, dir string, long, all, humanReadable bool) {
dirReader, err := client.Open(dir)
if err != nil {
fatal(err)
}
var tw *tabwriter.Writer
if long {
tw = lsTabWriter()
defer tw.Flush()
}
if all {
if long {
dirInfo, err := client.Stat(dir)
if err != nil {
fatal(err)
}
parentPath := path.Join(dir, "..")
parentInfo, err := client.Stat(parentPath)
if err != nil {
fatal(err)
}
printLong(tw, ".", dirInfo, humanReadable)
printLong(tw, "..", parentInfo, humanReadable)
} else {
fmt.Println(".")
fmt.Println("..")
}
}
var partial []os.FileInfo
for ; err != io.EOF; partial, err = dirReader.Readdir(100) {
if err != nil {
fatal(err)
}
printFiles(tw, partial, long, all, humanReadable)
}
if long {
tw.Flush()
}
}
