如何从 a.gz 存档中获取原始文件的名称？

Question

I'm writing a utility that takes a.gz archive and checks if it's contents already exist in a specified folder. If they don't, it will extract the archive there.

The way I planned on doing this is reading the filenames of the files in the.gz archive one by one and checking if such a file already exists in my directory. But from what I understand this isn't possible with gzip.

Ideally, I'm looking for something like this:

archive = gzipfile.GzipFile(source)

    for i in archive.getmembers():
        if os.path.isfile(destination + sep + i.name) and overwrite:
        ...

Is this possible?

Answer 1

A .gz file is not an archive, it simply is compressed. If you have a .tar.gz file, you can use tarfile .

Answer 2

While it is true that the .gz file is simply a compressed file, the original file name can be truncated or indeed the compressed .gz file can be renamed. gunzip can be told to provide the original file name using the -N flag and if used with the -l (minus lowercase L) it will tell you the original file name without uncompressing the file.
For example:

$ gzip sometext.txt
$ mv sometext.txt.gz othertext.gz
$ gunzip -Nl othertext.gz
         compressed        uncompressed  ratio uncompressed_name
                 58                 113  76.1% sometext.txt

You can hack your way through this in python as well.

from subprocess import check_output
size_name = check_output(['gunzip', '-Nlq','othertext.gz'])
size_name = size_name.strip().split("%",1)
print "original filename =",size_name[1].strip()

result:

original filename = sometext.txt

I do not believe that the python gzip package allows you to access the original file name.
Someone else may know different!

Answer 3

Adding to the accepted answer:

At least CPython's gzip does not expose the file name metadata because it simply discards it as you can see when you check the source code.

However, the gzip file format (specified in RFC 1952 ) or at least its metadata is easy enough to manually parse:

import struct

def getGzipName(path):
    with open(path, 'rb') as file:
        id1, id2, compression, flags, mtime, extraFlags, osId = struct.unpack('<BBBBLBB', file.read(10))
        if id1 != 0x1F or id2 != 0x8B or compression != 0x08:
            return None

        # Extra Field (e.g. used by bgzip to store the length of the compressed block)
        if flags & ( 1 << 2 ) != 0:
            file.read(struct.unpack('<U', file.read(2))[0])

        # File Name Field
        if flags & ( 1 << 3 ) != 0:
            name = b''
            c = file.read(1)
            while c != b'\0':
                name += c
                c = file.read(1)
            return name.decode()

    return None

Note that theoretically gzip could be used as an archive format because it does support storing original file names, which might be used to store paths, and because multiple gzip streams (all with different file names) are allowed to be concatenated to each other. However, not even the gzip tool does support such exotic gzip files not even with the --name option. It will simply concatenate the data of the second gzip stream to the original filename of the first gzip stream.

Answer 4

import tarfile

archive = tarfile.open(source)
for i in archive.getmembers():
    ...