hardlink_rpms

#!/usr/bin/python

import sys
import os
from optparse import OptionParser
import hashlib

"""
Compare the rpms in the given directories, if they are the same by chksum, hardlink them
build up a hash of seen rpms and chksum's, compare and link if equal.
extra credit - when not equal store the new hash and location into a new entry.
"""
global hashSums
global options
global count
global space
hashSums = {}
count = 0
space = 0


def human_size(size,precision=2):
  suffixes=['B','KB','MB','GB','TB']
  suffixIndex = 0
  while size > 1024:
    suffixIndex += 1
    size = size/1024.0
  return "%.*f%s" % (precision,size,suffixes[suffixIndex])

def chksum_file(path):
  """compute chksum for file, insert into hash"""
  global options
  if options.debug: 
	print(" -> computing chksum for %s" % path)
  try:
    hash = hashlib.sha1(open(path).read()) 
  except:
    print "Error: problem hashing %s" % path
    return -1
  return hash.hexdigest()

def find_file(filename,path,chksumX):
  """look for file in hash, return chksum,path tuple"""
  global options
  global hashSums

  try:
	(pathY,chksumY) = hashSums[filename]
	if options.debug: 
		print( " -> found %s at %s" % (filename,pathY))
	return (pathY,chksumY)
  except:
	hashSums[filename] = (path,chksumX)
	if options.debug: 
		print(" -> storing %s in hashSums" % filename)
	return (-1,-1)

def walk_dir(path):
  """look for all rpms in a path, check if they can be hardlinked along the way"""
  for root,dirs,files in os.walk(path):
	for dir in dirs:
		walk_dir(dir)
	for file in files:
		# file should be rpm
		if file[len(file)-4:] == '.rpm':
			check_file(file,root)

def hardlink(filename,path1,path2):
  """hard link path2 to path1, remove path2 first, then hardlink"""
  global options
  global count
  global space
  if options.debug: 
    print "linking %s in %s to %s" % (filename,path1,path2)
  path1 = "%s/%s" % (path1,filename)
  path2 = "%s/%s" % (path2,filename)
  try:
    inodeX = os.stat(path1).st_ino
    inodeY = os.stat(path2).st_ino
  except:
    print "Error: problem getting inode for %s or %s" % (path1,path2)
    return -1
  if inodeX != inodeY:
    try:
      os.unlink(path2)
      os.link(path1,path2)
      count = count + 1
      space = space + os.stat(path2).st_size
    except:
      print "Error: could not link %s to %s" % (path1,path2)
      return -1
  return 0
	


def check_file(filename,path):
  """get chksum for the file, look for it in hash, hardlink if appropriate"""
  chksumX = chksum_file("%s/%s" % (path,filename))
  (pathY,chksumY) = find_file(filename,path,chksumX)
  if chksumX == chksumY:
	hardlink(filename,path,pathY)

if __name__ == "__main__":
  usage = """usage: %prog [options] path [path2 path3 ...]\nwalk through path looking for rpms that can be hardlinked to one another"""
  parser = OptionParser(usage=usage)
  parser.add_option('-v','--verbose',dest='verbose',help='verbose output',action='store_true')
  parser.add_option('-d','--debug',dest='debug',help='debugging output',action='store_true')
  (options,args) = parser.parse_args()
  
  if not args:
    parser.error('at least one path must be given')
  if options.verbose: print "hardlinking rpms in ",
  for path in args:
    if options.verbose: print path +" ",
    walk_dir(path)
  if options.verbose:
    print "done.\nlinked %s rpms saving %s." % (count,human_size(space))