1 | #!/usr/bin/env 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 | # The 'gzip' module in the standard library provides a more complete
|
---|
6 | # implementation of gzip-format files.
|
---|
7 |
|
---|
8 | import zlib, sys, os
|
---|
9 |
|
---|
10 | FTEXT, FHCRC, FEXTRA, FNAME, FCOMMENT = 1, 2, 4, 8, 16
|
---|
11 |
|
---|
12 | def write32(output, value):
|
---|
13 | output.write(chr(value & 255)) ; value=value // 256
|
---|
14 | output.write(chr(value & 255)) ; value=value // 256
|
---|
15 | output.write(chr(value & 255)) ; value=value // 256
|
---|
16 | output.write(chr(value & 255))
|
---|
17 |
|
---|
18 | def read32(input):
|
---|
19 | v = ord(input.read(1))
|
---|
20 | v += (ord(input.read(1)) << 8 )
|
---|
21 | v += (ord(input.read(1)) << 16)
|
---|
22 | v += (ord(input.read(1)) << 24)
|
---|
23 | return v
|
---|
24 |
|
---|
25 | def compress (filename, input, output):
|
---|
26 | output.write('\037\213\010') # Write the header, ...
|
---|
27 | output.write(chr(FNAME)) # ... flag byte ...
|
---|
28 |
|
---|
29 | statval = os.stat(filename) # ... modification time ...
|
---|
30 | mtime = statval[8]
|
---|
31 | write32(output, mtime)
|
---|
32 | output.write('\002') # ... slowest compression alg. ...
|
---|
33 | output.write('\377') # ... OS (=unknown) ...
|
---|
34 | output.write(filename+'\000') # ... original filename ...
|
---|
35 |
|
---|
36 | crcval = zlib.crc32("")
|
---|
37 | compobj = zlib.compressobj(9, zlib.DEFLATED, -zlib.MAX_WBITS,
|
---|
38 | zlib.DEF_MEM_LEVEL, 0)
|
---|
39 | while True:
|
---|
40 | data = input.read(1024)
|
---|
41 | if data == "":
|
---|
42 | break
|
---|
43 | crcval = zlib.crc32(data, crcval)
|
---|
44 | output.write(compobj.compress(data))
|
---|
45 | output.write(compobj.flush())
|
---|
46 | write32(output, crcval) # ... the CRC ...
|
---|
47 | write32(output, statval[6]) # and the file size.
|
---|
48 |
|
---|
49 | def decompress (input, output):
|
---|
50 | magic = input.read(2)
|
---|
51 | if magic != '\037\213':
|
---|
52 | print 'Not a gzipped file'
|
---|
53 | sys.exit(0)
|
---|
54 | if ord(input.read(1)) != 8:
|
---|
55 | print 'Unknown compression method'
|
---|
56 | sys.exit(0)
|
---|
57 | flag = ord(input.read(1))
|
---|
58 | input.read(4+1+1) # Discard modification time,
|
---|
59 | # extra flags, and OS byte.
|
---|
60 | if flag & FEXTRA:
|
---|
61 | # Read & discard the extra field, if present
|
---|
62 | xlen = ord(input.read(1))
|
---|
63 | xlen += 256*ord(input.read(1))
|
---|
64 | input.read(xlen)
|
---|
65 | if flag & FNAME:
|
---|
66 | # Read and discard a null-terminated string containing the filename
|
---|
67 | while True:
|
---|
68 | s = input.read(1)
|
---|
69 | if s == '\0': break
|
---|
70 | if flag & FCOMMENT:
|
---|
71 | # Read and discard a null-terminated string containing a comment
|
---|
72 | while True:
|
---|
73 | s=input.read(1)
|
---|
74 | if s=='\0': break
|
---|
75 | if flag & FHCRC:
|
---|
76 | input.read(2) # Read & discard the 16-bit header CRC
|
---|
77 |
|
---|
78 | decompobj = zlib.decompressobj(-zlib.MAX_WBITS)
|
---|
79 | crcval = zlib.crc32("")
|
---|
80 | length = 0
|
---|
81 | while True:
|
---|
82 | data=input.read(1024)
|
---|
83 | if data == "":
|
---|
84 | break
|
---|
85 | decompdata = decompobj.decompress(data)
|
---|
86 | output.write(decompdata)
|
---|
87 | length += len(decompdata)
|
---|
88 | crcval = zlib.crc32(decompdata, crcval)
|
---|
89 |
|
---|
90 | decompdata = decompobj.flush()
|
---|
91 | output.write(decompdata)
|
---|
92 | length += len(decompdata)
|
---|
93 | crcval = zlib.crc32(decompdata, crcval)
|
---|
94 |
|
---|
95 | # We've read to the end of the file, so we have to rewind in order
|
---|
96 | # to reread the 8 bytes containing the CRC and the file size. The
|
---|
97 | # decompressor is smart and knows when to stop, so feeding it
|
---|
98 | # extra data is harmless.
|
---|
99 | input.seek(-8, 2)
|
---|
100 | crc32 = read32(input)
|
---|
101 | isize = read32(input)
|
---|
102 | if crc32 != crcval:
|
---|
103 | print 'CRC check failed.'
|
---|
104 | if isize != length:
|
---|
105 | print 'Incorrect length of data produced'
|
---|
106 |
|
---|
107 | def main():
|
---|
108 | if len(sys.argv)!=2:
|
---|
109 | print 'Usage: minigzip.py <filename>'
|
---|
110 | print ' The file will be compressed or decompressed.'
|
---|
111 | sys.exit(0)
|
---|
112 |
|
---|
113 | filename = sys.argv[1]
|
---|
114 | if filename.endswith('.gz'):
|
---|
115 | compressing = False
|
---|
116 | outputname = filename[:-3]
|
---|
117 | else:
|
---|
118 | compressing = True
|
---|
119 | outputname = filename + '.gz'
|
---|
120 |
|
---|
121 | input = open(filename, 'rb')
|
---|
122 | output = open(outputname, 'wb')
|
---|
123 |
|
---|
124 | if compressing:
|
---|
125 | compress(filename, input, output)
|
---|
126 | else:
|
---|
127 | decompress(input, output)
|
---|
128 |
|
---|
129 | input.close()
|
---|
130 | output.close()
|
---|
131 |
|
---|
132 | if __name__ == '__main__':
|
---|
133 | main()
|
---|