dnet-hadoop/dhp-common/src/main/java/eu/dnetlib/dhp/common/MakeTarArchive.java

115 lines
3.4 KiB
Java
Raw Normal View History

2020-11-04 18:12:43 +01:00
package eu.dnetlib.dhp.common;
import java.io.BufferedInputStream;
import java.io.IOException;
import java.io.InputStream;
import java.io.Serializable;
2020-11-04 18:12:43 +01:00
import org.apache.commons.compress.archivers.tar.TarArchiveEntry;
import org.apache.commons.compress.archivers.tar.TarArchiveOutputStream;
import org.apache.hadoop.fs.*;
2020-11-04 18:12:43 +01:00
public class MakeTarArchive implements Serializable {
2020-11-04 18:12:43 +01:00
private static TarArchiveOutputStream getTar(FileSystem fileSystem, String outputPath) throws IOException {
Path hdfsWritePath = new Path(outputPath);
FSDataOutputStream fsDataOutputStream = null;
if (fileSystem.exists(hdfsWritePath)) {
fileSystem.delete(hdfsWritePath, true);
2020-11-04 18:12:43 +01:00
}
fsDataOutputStream = fileSystem.create(hdfsWritePath);
2020-11-04 18:12:43 +01:00
return new TarArchiveOutputStream(fsDataOutputStream.getWrappedStream());
}
2020-11-04 18:12:43 +01:00
private static void write(FileSystem fileSystem, String inputPath, String outputPath, String dir_name)
throws IOException {
2020-11-04 18:12:43 +01:00
Path hdfsWritePath = new Path(outputPath);
FSDataOutputStream fsDataOutputStream = null;
if (fileSystem.exists(hdfsWritePath)) {
fileSystem.delete(hdfsWritePath, true);
2020-11-04 18:12:43 +01:00
}
fsDataOutputStream = fileSystem.create(hdfsWritePath);
2020-11-04 18:12:43 +01:00
TarArchiveOutputStream ar = new TarArchiveOutputStream(fsDataOutputStream.getWrappedStream());
2020-11-04 18:12:43 +01:00
RemoteIterator<LocatedFileStatus> fileStatusListIterator = fileSystem
.listFiles(
new Path(inputPath), true);
2020-11-04 18:12:43 +01:00
while (fileStatusListIterator.hasNext()) {
writeCurrentFile(fileSystem, dir_name, fileStatusListIterator, ar, 0);
}
2020-11-04 18:12:43 +01:00
ar.close();
}
2020-11-04 18:12:43 +01:00
public static void tarMaxSize(FileSystem fileSystem, String inputPath, String outputPath, String dir_name,
int gBperSplit) throws IOException {
final long bytesPerSplit = 1024L * 1024L * 1024L * gBperSplit;
2020-11-04 18:12:43 +01:00
long sourceSize = fileSystem.getContentSummary(new Path(inputPath)).getSpaceConsumed();
2020-11-04 18:12:43 +01:00
if (sourceSize < bytesPerSplit) {
write(fileSystem, inputPath, outputPath + ".tar", dir_name);
} else {
int partNum = 0;
2020-11-04 18:12:43 +01:00
RemoteIterator<LocatedFileStatus> fileStatusListIterator = fileSystem
.listFiles(
new Path(inputPath), true);
boolean next = fileStatusListIterator.hasNext();
while (next) {
TarArchiveOutputStream ar = getTar(fileSystem, outputPath + "_" + (partNum + 1) + ".tar");
2020-11-04 18:12:43 +01:00
long current_size = 0;
while (next && current_size < bytesPerSplit) {
current_size = writeCurrentFile(fileSystem, dir_name, fileStatusListIterator, ar, current_size);
next = fileStatusListIterator.hasNext();
2020-11-04 18:12:43 +01:00
}
2020-11-04 18:12:43 +01:00
partNum += 1;
ar.close();
}
2020-11-04 18:12:43 +01:00
}
2020-11-04 18:12:43 +01:00
}
2020-11-04 18:12:43 +01:00
private static long writeCurrentFile(FileSystem fileSystem, String dir_name,
RemoteIterator<LocatedFileStatus> fileStatusListIterator,
TarArchiveOutputStream ar, long current_size) throws IOException {
LocatedFileStatus fileStatus = fileStatusListIterator.next();
2020-11-04 18:12:43 +01:00
Path p = fileStatus.getPath();
String p_string = p.toString();
if (!p_string.endsWith("_SUCCESS")) {
String name = p_string.substring(p_string.lastIndexOf("/") + 1);
TarArchiveEntry entry = new TarArchiveEntry(dir_name + "/" + name);
entry.setSize(fileStatus.getLen());
current_size += fileStatus.getLen();
ar.putArchiveEntry(entry);
2020-11-04 18:12:43 +01:00
InputStream is = fileSystem.open(fileStatus.getPath());
2020-11-04 18:12:43 +01:00
BufferedInputStream bis = new BufferedInputStream(is);
2020-11-04 18:12:43 +01:00
int count;
byte data[] = new byte[1024];
while ((count = bis.read(data, 0, data.length)) != -1) {
ar.write(data, 0, count);
}
bis.close();
ar.closeArchiveEntry();
2020-11-04 18:12:43 +01:00
}
return current_size;
}
}