Is there a Java equivalent of GetCompressedFileSize?

Question

I am looking to get accurate (i.e. the real size on disk and not the normal size that includes all the 0\'s) measurements of sparse files in Java.

In C++ on Windows one

Answer 1

If you want a pure Java solution you can try jnr-posix. Here's an example implementation

import jnr.posix.*;

final POSIX p = POSIXFactory.getPOSIX();
final int S_BLKSIZE = 512; // from sys/stat.h
final FileStat stat = p.stat("/path/to/file");
final long bytes = stat.blocks() * S_BLKSIZE;

However currently the function won't work for Windows. Until that's fixed you have to use platform-specific code like below

• On Linux use the stat64 system call

  The st_blocks field indicates the number of blocks allocated to the file, 512-byte units. (This may be smaller than st_size/512 when the file has holes.)

  • You can also run the stat command. The number of allocated blocks can be seen in the Blocks field, or printed with the %b format specifier
  • Or use du command (without --apparent-size option)

    --apparent-size

    • print apparent sizes, rather than disk usage; although the apparent size is usually smaller, it may be larger due to holes in ('sparse') files, internal fragmentation, indirect blocks, and the like

• On Windows you can call the GetCompressedFileSize API

  • Alternatively you can also run fsutil file layout with admin rights to get detailed information about a file. Find the $DATA stream.

    • If you see Resident | No clusters allocated in the flags like this then it's a resident file and size on disk would be 0.

      PS C:\Users>  fsutil file layout .\desktop.ini
      
      ********* File 0x000800000003dbde *********
      File reference number   : 0x000800000003dbde
      File attributes         : 0x00000026: Hidden | System | Archive
      File entry flags        : 0x00000000
      Link (ParentID: Name)   : 0x001f0000000238c8: HLINK Name   : \Users\desktop.ini
      ...
      Stream                  : 0x080  ::$DATA
          Attributes          : 0x00000000: *NONE*
          Flags               : 0x0000000c: Resident | No clusters allocated
          Size                : 174
          Allocated Size      : 176
      

    • If you don't see the resident flag then check the Allocated Size field, it's the file's size on disk

      PS D:\>  fsutil file layout .\nonresident.txt
      
      ********* File 0x000400000000084e *********
      File reference number   : 0x000400000000084e
      File attributes         : 0x00000020: Archive
      File entry flags        : 0x00000000
      Link (ParentID: Name)   : 0x0005000000000005: HLINK Name   : \nonresident.txt
      ...
      Stream                  : 0x080  ::$DATA
          Attributes          : 0x00000000: *NONE*
          Flags               : 0x00000000: *NONE*
          Size                : 1,520
          Allocated Size      : 4,096
          Extents             : 1 Extents
                              : 1: VCN: 0 Clusters: 1 LCN: 1,497,204
      

For more information you can read the below questions

• How do I query "Size on disk" file information?
• Get size of file on disk

Answer 2

Since an answer was given for windows. i will try to supply for Linux.

I am not sure, but i think it will do the trick (C++):

#include <linux/fs.h>
ioctl(file, BLKGETSIZE64, &file_size_in_bytes);

This can be loaded in the same way that was described in the @Aniket answer (JNI)

Answer 3

If you are doing it on Windows alone, you can write it with Java Native Interface

class NativeInterface{
   public static native long GetCompressedFileSize(String filename);
}

and in C/C++ file:

extern "C"
JNIEXPORT jlong JNICALL Java_NativeInterface_GetCompressedFileSize
  (JNIEnv *env, jobject obj, jstring javaString)
{
    const char *nativeString = env->GetStringUTFChars(javaString, 0);

    char buffer[512];
    strcpy(buffer, nativeString);
    env->ReleaseStringUTFChars(javaString, nativeString);
    return (jlong) GetCompressedFileSize(buffer, NULL);
}