Demo/zlib/minigzip.py

   1 #!/usr/local/bin/python
   2 # Demo program for zlib; it compresses or decompresses files, but *doesn't*
   3 # delete the original.  This doesn't support all of gzip's options.
   4
   5 FTEXT, FHCRC, FEXTRA, FNAME, FCOMMENT = 1, 2, 4, 8, 16
   6
   7 def write32(output, value):
   8     output.write(chr(value & 255)) ; value=value / 256
   9     output.write(chr(value & 255)) ; value=value / 256
  10     output.write(chr(value & 255)) ; value=value / 256
  11     output.write(chr(value & 255))
  12
  13 def read32(input):
  14     v=ord(input.read(1))
  15     v=v+ (ord(input.read(1))<<8 )
  16     v=v+ (ord(input.read(1))<<16)
  17     v=v+ (ord(input.read(1))<<24)
  18     return v
  19
  20 import zlib, sys
  21 if len(sys.argv)!=2:
  22    print 'Usage: minigzip.py <filename>'
  23    print '  The file will be compressed or decompressed.'
  24    sys.exit(0)
  25
  26 filename=sys.argv[1]
  27 compressing=1 ; outputname=filename+'.gz'
  28 if filename[-3:]=='.gz':
  29     compressing=0 ; outputname=filename[:-3]
  30 input=open(filename) ; output=open(outputname, 'w')
  31
  32 if compressing:
  33     output.write('\037\213\010')        # Write the header, ...
  34     output.write(chr(FNAME))            # ... flag byte ...
  35
  36     import os                           # ... modification time ...
  37     statval=os.stat(filename)
  38     mtime=statval[8]
  39     write32(output, mtime)
  40     output.write('\002')                # ... slowest compression alg. ...
  41     output.write('\377')                # ... OS (=unknown) ...
  42     output.write(filename+'\000')       # ... original filename ...
  43
  44     crcval=zlib.crc32("")
  45     compobj=zlib.compressobj(9, zlib.DEFLATED, -zlib.MAX_WBITS,
  46                              zlib.DEF_MEM_LEVEL, 0)
  47     while (1):
  48         data=input.read(1024)
  49         if data=="": break
  50         crcval=zlib.crc32(data, crcval)
  51         output.write(compobj.compress(data))
  52     output.write(compobj.flush())
  53     write32(output, crcval)             # ... the CRC ...
  54     write32(output, statval[6])         # and the file size.
  55
  56 else:
  57     magic=input.read(2)
  58     if magic!='\037\213':
  59         print 'Not a gzipped file' ; sys.exit(0)
  60     if ord(input.read(1))!=8:
  61         print 'Unknown compression method' ; sys.exit(0)
  62     flag=ord(input.read(1))
  63     input.read(4+1+1)                   # Discard modification time,
  64                                         # extra flags, and OS byte.
  65     if flag & FEXTRA:
  66         # Read & discard the extra field, if present
  67         xlen=ord(input.read(1))
  68         xlen=xlen+256*ord(input.read(1))
  69         input.read(xlen)
  70     if flag & FNAME:
  71         # Read and discard a null-terminated string containing the filename
  72         while (1):
  73             s=input.read(1)
  74             if s=='\000': break
  75     if flag & FCOMMENT:
  76         # Read and discard a null-terminated string containing a comment
  77         while (1):
  78             s=input.read(1)
  79             if s=='\000': break
  80     if flag & FHCRC:
  81         input.read(2)                   # Read & discard the 16-bit header CRC
  82     decompobj=zlib.decompressobj(-zlib.MAX_WBITS)
  83     crcval=zlib.crc32("")
  84     length=0
  85     while (1):
  86         data=input.read(1024)
  87         if data=="": break
  88         decompdata=decompobj.decompress(data)
  89         print len(decompdata)
  90         output.write(decompdata) ; length=length+len(decompdata)
  91         crcval=zlib.crc32(decompdata, crcval)
  92     decompdata=decompobj.flush()
  93     output.write(decompdata) ; length=length+len(decompdata)
  94     crcval=zlib.crc32(decompdata, crcval)
  95
  96     # We've read to the end of the file, so we have to rewind in order
  97     # to reread the 8 bytes containing the CRC and the file size.  The
  98     # decompressor is smart and knows when to stop, so feeding it
  99     # extra data is harmless.
 100     input.seek(-8, 2)
 101     crc32=read32(input)
 102     isize=read32(input)
 103     if crc32!=crcval: print 'CRC check failed.'
 104     if isize!=length: print 'Incorrect length of data produced'
 105
 106 input.close() ; output.close()
 107