forked from D-Net/dnet-hadoop
removed the part after part-x- in the file name generated by spark. It was too long and created problems while creating the tar entries
This commit is contained in:
parent
618d2de2da
commit
d418c309f5
|
@ -90,6 +90,13 @@ public class MakeTarArchive implements Serializable {
|
|||
String p_string = p.toString();
|
||||
if (!p_string.endsWith("_SUCCESS")) {
|
||||
String name = p_string.substring(p_string.lastIndexOf("/") + 1);
|
||||
if (name.startsWith("part-") & name.length() > 10) {
|
||||
String tmp = name.substring(0, 10);
|
||||
if (name.contains(".")) {
|
||||
tmp += name.substring(name.indexOf("."));
|
||||
}
|
||||
name = tmp;
|
||||
}
|
||||
TarArchiveEntry entry = new TarArchiveEntry(dir_name + "/" + name);
|
||||
entry.setSize(fileStatus.getLen());
|
||||
current_size += fileStatus.getLen();
|
||||
|
|
Loading…
Reference in New Issue