1  #! /usr/bin/env python3
       2  
       3  """Python utility to print MD5 checksums of argument files.
       4  """
       5  
       6  
       7  bufsize = 8096
       8  fnfilter = None
       9  rmode = 'rb'
      10  
      11  usage = """
      12  usage: md5sum.py [-b] [-t] [-l] [-s bufsize] [file ...]
      13  -b        : read files in binary mode (default)
      14  -t        : read files in text mode (you almost certainly don't want this!)
      15  -l        : print last pathname component only
      16  -s bufsize: read buffer size (default %d)
      17  file ...  : files to sum; '-' or no files means stdin
      18  """ % bufsize
      19  
      20  import io
      21  import sys
      22  import os
      23  import getopt
      24  from hashlib import md5
      25  
      26  def sum(*files):
      27      sts = 0
      28      if files and isinstance(files[-1], io.IOBase):
      29          out, files = files[-1], files[:-1]
      30      else:
      31          out = sys.stdout
      32      if len(files) == 1 and not isinstance(files[0], str):
      33          files = files[0]
      34      for f in files:
      35          if isinstance(f, str):
      36              if f == '-':
      37                  sts = printsumfp(sys.stdin, '<stdin>', out) or sts
      38              else:
      39                  sts = printsum(f, out) or sts
      40          else:
      41              sts = sum(f, out) or sts
      42      return sts
      43  
      44  def printsum(filename, out=sys.stdout):
      45      try:
      46          fp = open(filename, rmode)
      47      except IOError as msg:
      48          sys.stderr.write('%s: Can\'t open: %s\n' % (filename, msg))
      49          return 1
      50      with fp:
      51          if fnfilter:
      52              filename = fnfilter(filename)
      53          sts = printsumfp(fp, filename, out)
      54      return sts
      55  
      56  def printsumfp(fp, filename, out=sys.stdout):
      57      m = md5()
      58      try:
      59          while 1:
      60              data = fp.read(bufsize)
      61              if not data:
      62                  break
      63              if isinstance(data, str):
      64                  data = data.encode(fp.encoding)
      65              m.update(data)
      66      except IOError as msg:
      67          sys.stderr.write('%s: I/O error: %s\n' % (filename, msg))
      68          return 1
      69      out.write('%s %s\n' % (m.hexdigest(), filename))
      70      return 0
      71  
      72  def main(args = sys.argv[1:], out=sys.stdout):
      73      global fnfilter, rmode, bufsize
      74      try:
      75          opts, args = getopt.getopt(args, 'blts:')
      76      except getopt.error as msg:
      77          sys.stderr.write('%s: %s\n%s' % (sys.argv[0], msg, usage))
      78          return 2
      79      for o, a in opts:
      80          if o == '-l':
      81              fnfilter = os.path.basename
      82          elif o == '-b':
      83              rmode = 'rb'
      84          elif o == '-t':
      85              rmode = 'r'
      86          elif o == '-s':
      87              bufsize = int(a)
      88      if not args:
      89          args = ['-']
      90      return sum(args, out)
      91  
      92  if __name__ == '__main__' or __name__ == sys.argv[0]:
      93      sys.exit(main(sys.argv[1:], sys.stdout))