file-largest

#!/usr/bin/python3

import os
import sys


def kmg(bytes,kilo=1000, append='',thresh=15,rstrip0=1, extradigits=0):
    """ Readable size formatter
        e.g. '%sB'%kmg(2342342324) = '2.3 GB'
             kmg(3429873278462) == '3.1T'
             kmg(342987327)     == '327M'
             kmg(34298)         == '33K'
        Decimal/SI kilos by default. Specify kilo=1024 if you want binary kilos.

        Maybe use sfloat-like behaviour for the number?
    """
    ret = None
    
    mega = kilo*kilo
    giga = mega*kilo
    tera = giga*kilo
    peta = tera*kilo

    # Yes, can be abstracted now
    if abs(bytes)>(0.80*peta):
        showval = bytes/float(peta)
        if showval<thresh:
            showdigits = 1 + extradigits
        else:
            showdigits = 0 + extradigits
        append+='P'

    elif abs(bytes)>(0.80*tera):
        showval = bytes/float(tera)
        if showval<thresh: 
            showdigits = 1 + extradigits
        else:
            showdigits = 0 + extradigits
        append+='T'

    elif abs(bytes)>(0.95*giga):
        showval = bytes/float(giga)
        if showval<thresh: # e.g. 1.3GB but 15GB
            showdigits = 1 + extradigits
        else:
            showdigits = 0 + extradigits
        append+='G'
            
    elif abs(bytes)>(0.9*mega):
        showval = bytes/float(mega)
        if showval<thresh:
            showdigits = 1 + extradigits
        else:
            showdigits = 0 + extradigits
        append+='M'
    
    elif abs(bytes)>(0.85*kilo):
        showval = bytes/float(kilo)
        if showval<thresh:
            showdigits = 1 + extradigits
        else:
            showdigits = 0 + extradigits
        append+='K'
    else:
        showval = bytes
        showdigits=0

    ret = ("%%.%df"%(showdigits))%showval
        
    if rstrip0:
        if '.' in ret:
            ret=ret.rstrip('0').rstrip('.')                   

    ret+=append
        
    return ret


data = [] # list of (size, filename)

print( "Reading...", end='')
cnt=0
for path in sys.argv[1:]:
    for r,ds,fs in os.walk(path):
    	for fn in fs:
            ffn = os.path.join( r, fn )
            mode, inode, device, numhardlinks, uid,gid, size, atime, mtime, ctime = os.lstat(ffn)
            data.append( (size, ffn) )
            cnt += 1
            if cnt%1000==0:
                sys.stdout.write('%s '%cnt)
                sys.stdout.flush()
print( "Done" )    

# TODO: option parsing
show_amt = 30
            
for size, fn in sorted( data )[-show_amt:]:
    print( '%6sB\t%r'%( kmg(size), fn ))