file-detectsparse

#!/usr/bin/python3

''' Try to figure out whether a file is sparese, and if so, report how much of it is used.

    based on code at https://stackoverflow.com/questions/46100881/sparse-files-how-to-find-contents

    Note that SEEK_END / SEEK_HOLE have a history and cannot be considered fully predictable
    https://lwn.net/Articles/440255/
'''


import os
import sys
import errno
import helpers_format
import json

SEEK_DATA = 3
SEEK_HOLE = 4

def get_ranges(fobj):
    '''
        For a possibly-sparse file, find the actually-allocated parts.
        Returns:
        - sum of used ranges
        - list of (start, end) of allocated blocks
    '''
    fileno = fobj.fileno()
    ranges = []
    end = 0

    sum_used = 0
    while True:
        try:
            start = os.lseek( fileno, end,   SEEK_DATA)
            end   = os.lseek( fileno, start, SEEK_HOLE)
            ranges.append((start, end))
            sum_used += end-start
        except OSError as e:
            if e.errno == errno.ENXIO: # for this operation, it effectively means 'done with file'?
                return sum_used, ranges
            raise

def main():
    for fn in sys.argv[1:]:
        # various seeks are invalid for e.g. character files, so we already filter for regular fieles
        if not os.path.isfile(fn):
            continue

        if not os.path.exists(fn):
            continue
        
        #if os.path.islink(fn) and not os.path.exists(fn):
        #    print( "SKIP, Broken symlink: %r"%fn )
        #    continue
        
        #print("CHECKING ",fn)
        try:
            with open(fn, 'rb') as f:
                try:
                    fileno = f.fileno()
                    os.lseek(fileno, 0, os.SEEK_END)
                    filesize = f.tell() # mixing APIs, but it should be fine
                    os.lseek(fileno, 0, os.SEEK_SET)
                except OSError as e:  # hm? special file?
                    #print( f'ERR1: {e} for {fn}' ) # skip for now, but TODO: figure out
                    continue
                
                sum_used, ranges = get_ranges(f)
                if filesize == 0:
                    use_frac = 1
                else:
                    use_frac = sum_used / filesize

                if use_frac == 1:
                    continue
                    
                print( '%10dK used of %10dK (%3.0f%%)  for %r'%( sum_used/1024, filesize/1024,  100. * use_frac, fn ) )
                
                #for start, end in ranges:
                #    print('[%d:%d]' % (start, end))
                #    size = end-start
                #    length = min(20, size)
                ##    f.seek(start)
                ##    data = f.read(length)
                ##    print(data)
        except OSError as e:
            print( f'ERR2: {e} for {fn}' )
            raise
            #raise SystemExit(1)


if __name__ == '__main__':
    main()