code formatting
parent
f5f346dd2b
commit
4625b7486e
@ -1,119 +1,117 @@
|
||||
package eu.dnetlib.dhp.common;
|
||||
|
||||
import org.apache.commons.compress.archivers.tar.TarArchiveEntry;
|
||||
import org.apache.commons.compress.archivers.tar.TarArchiveOutputStream;
|
||||
import org.apache.hadoop.fs.*;
|
||||
package eu.dnetlib.dhp.common;
|
||||
|
||||
import java.io.BufferedInputStream;
|
||||
import java.io.IOException;
|
||||
import java.io.InputStream;
|
||||
import java.io.Serializable;
|
||||
|
||||
public class MakeTarArchive implements Serializable {
|
||||
|
||||
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);
|
||||
|
||||
}
|
||||
fsDataOutputStream = fileSystem.create(hdfsWritePath);
|
||||
import org.apache.commons.compress.archivers.tar.TarArchiveEntry;
|
||||
import org.apache.commons.compress.archivers.tar.TarArchiveOutputStream;
|
||||
import org.apache.hadoop.fs.*;
|
||||
|
||||
return new TarArchiveOutputStream(fsDataOutputStream.getWrappedStream());
|
||||
}
|
||||
public class MakeTarArchive implements Serializable {
|
||||
|
||||
private static void write(FileSystem fileSystem, String inputPath, String outputPath, String dir_name)
|
||||
throws IOException {
|
||||
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);
|
||||
|
||||
Path hdfsWritePath = new Path(outputPath);
|
||||
FSDataOutputStream fsDataOutputStream = null;
|
||||
if (fileSystem.exists(hdfsWritePath)) {
|
||||
fileSystem.delete(hdfsWritePath, true);
|
||||
}
|
||||
fsDataOutputStream = fileSystem.create(hdfsWritePath);
|
||||
|
||||
}
|
||||
fsDataOutputStream = fileSystem.create(hdfsWritePath);
|
||||
return new TarArchiveOutputStream(fsDataOutputStream.getWrappedStream());
|
||||
}
|
||||
|
||||
TarArchiveOutputStream ar = new TarArchiveOutputStream(fsDataOutputStream.getWrappedStream());
|
||||
private static void write(FileSystem fileSystem, String inputPath, String outputPath, String dir_name)
|
||||
throws IOException {
|
||||
|
||||
RemoteIterator<LocatedFileStatus> fileStatusListIterator = fileSystem
|
||||
.listFiles(
|
||||
new Path(inputPath), true);
|
||||
Path hdfsWritePath = new Path(outputPath);
|
||||
FSDataOutputStream fsDataOutputStream = null;
|
||||
if (fileSystem.exists(hdfsWritePath)) {
|
||||
fileSystem.delete(hdfsWritePath, true);
|
||||
|
||||
while (fileStatusListIterator.hasNext()) {
|
||||
writeCurrentFile(fileSystem, dir_name, fileStatusListIterator, ar, 0);
|
||||
}
|
||||
}
|
||||
fsDataOutputStream = fileSystem.create(hdfsWritePath);
|
||||
|
||||
ar.close();
|
||||
}
|
||||
TarArchiveOutputStream ar = new TarArchiveOutputStream(fsDataOutputStream.getWrappedStream());
|
||||
|
||||
public static void tarMaxSize(FileSystem fileSystem, String inputPath, String outputPath, String dir_name,
|
||||
int gBperSplit) throws IOException {
|
||||
final long bytesPerSplit = 1024L * 1024L * 1024L * gBperSplit;
|
||||
RemoteIterator<LocatedFileStatus> fileStatusListIterator = fileSystem
|
||||
.listFiles(
|
||||
new Path(inputPath), true);
|
||||
|
||||
long sourceSize = fileSystem.getContentSummary(new Path(inputPath)).getSpaceConsumed();
|
||||
while (fileStatusListIterator.hasNext()) {
|
||||
writeCurrentFile(fileSystem, dir_name, fileStatusListIterator, ar, 0);
|
||||
}
|
||||
|
||||
if (sourceSize < bytesPerSplit) {
|
||||
write(fileSystem, inputPath, outputPath + ".tar", dir_name);
|
||||
} else {
|
||||
int partNum = 0;
|
||||
ar.close();
|
||||
}
|
||||
|
||||
RemoteIterator<LocatedFileStatus> fileStatusListIterator = fileSystem
|
||||
.listFiles(
|
||||
new Path(inputPath), true);
|
||||
boolean next = fileStatusListIterator.hasNext();
|
||||
while (next) {
|
||||
TarArchiveOutputStream ar = getTar(fileSystem, outputPath + "_" + (partNum + 1) + ".tar");
|
||||
public static void tarMaxSize(FileSystem fileSystem, String inputPath, String outputPath, String dir_name,
|
||||
int gBperSplit) throws IOException {
|
||||
final long bytesPerSplit = 1024L * 1024L * 1024L * gBperSplit;
|
||||
|
||||
long current_size = 0;
|
||||
while (next && current_size < bytesPerSplit) {
|
||||
current_size = writeCurrentFile(fileSystem, dir_name, fileStatusListIterator, ar, current_size);
|
||||
next = fileStatusListIterator.hasNext();
|
||||
long sourceSize = fileSystem.getContentSummary(new Path(inputPath)).getSpaceConsumed();
|
||||
|
||||
}
|
||||
if (sourceSize < bytesPerSplit) {
|
||||
write(fileSystem, inputPath, outputPath + ".tar", dir_name);
|
||||
} else {
|
||||
int partNum = 0;
|
||||
|
||||
partNum += 1;
|
||||
ar.close();
|
||||
}
|
||||
RemoteIterator<LocatedFileStatus> fileStatusListIterator = fileSystem
|
||||
.listFiles(
|
||||
new Path(inputPath), true);
|
||||
boolean next = fileStatusListIterator.hasNext();
|
||||
while (next) {
|
||||
TarArchiveOutputStream ar = getTar(fileSystem, outputPath + "_" + (partNum + 1) + ".tar");
|
||||
|
||||
}
|
||||
long current_size = 0;
|
||||
while (next && current_size < bytesPerSplit) {
|
||||
current_size = writeCurrentFile(fileSystem, dir_name, fileStatusListIterator, ar, current_size);
|
||||
next = fileStatusListIterator.hasNext();
|
||||
|
||||
}
|
||||
}
|
||||
|
||||
private static long writeCurrentFile(FileSystem fileSystem, String dir_name,
|
||||
RemoteIterator<LocatedFileStatus> fileStatusListIterator,
|
||||
TarArchiveOutputStream ar, long current_size) throws IOException {
|
||||
LocatedFileStatus fileStatus = fileStatusListIterator.next();
|
||||
partNum += 1;
|
||||
ar.close();
|
||||
}
|
||||
|
||||
Path p = fileStatus.getPath();
|
||||
String p_string = p.toString();
|
||||
if (!p_string.endsWith("_SUCCESS")) {
|
||||
String name = p_string.substring(p_string.lastIndexOf("/") + 1);
|
||||
if (name.trim().equalsIgnoreCase("communities_infrastructures")) {
|
||||
name = "communities_infrastructures.json";
|
||||
}
|
||||
TarArchiveEntry entry = new TarArchiveEntry(dir_name + "/" + name);
|
||||
entry.setSize(fileStatus.getLen());
|
||||
current_size += fileStatus.getLen();
|
||||
ar.putArchiveEntry(entry);
|
||||
}
|
||||
|
||||
InputStream is = fileSystem.open(fileStatus.getPath());
|
||||
}
|
||||
|
||||
BufferedInputStream bis = new BufferedInputStream(is);
|
||||
private static long writeCurrentFile(FileSystem fileSystem, String dir_name,
|
||||
RemoteIterator<LocatedFileStatus> fileStatusListIterator,
|
||||
TarArchiveOutputStream ar, long current_size) throws IOException {
|
||||
LocatedFileStatus fileStatus = fileStatusListIterator.next();
|
||||
|
||||
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();
|
||||
Path p = fileStatus.getPath();
|
||||
String p_string = p.toString();
|
||||
if (!p_string.endsWith("_SUCCESS")) {
|
||||
String name = p_string.substring(p_string.lastIndexOf("/") + 1);
|
||||
if (name.trim().equalsIgnoreCase("communities_infrastructures")) {
|
||||
name = "communities_infrastructures.json";
|
||||
}
|
||||
TarArchiveEntry entry = new TarArchiveEntry(dir_name + "/" + name);
|
||||
entry.setSize(fileStatus.getLen());
|
||||
current_size += fileStatus.getLen();
|
||||
ar.putArchiveEntry(entry);
|
||||
|
||||
}
|
||||
return current_size;
|
||||
}
|
||||
InputStream is = fileSystem.open(fileStatus.getPath());
|
||||
|
||||
BufferedInputStream bis = new BufferedInputStream(is);
|
||||
|
||||
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();
|
||||
|
||||
}
|
||||
return current_size;
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
Loading…
Reference in New Issue