为什么文件内容没有复制到我的tarfile中

Question

Here's some code meant to copy the contents of a zipfile to a tarfile. I intend later to limit the copying to files that appear within a list that's passed in as a further argument, but for now, I'm just trying to get copying work.

import zipfile, tempfile, shutil, tarfile, os

def gather_and_repackage_files(zip_file_path, target_file_path) :
    with tarfile.open(target_file_path, "w") as tar:
        with zipfile.ZipFile(zip_file_path) as zip_file:
            for member in zip_file.namelist():
                filename = os.path.basename(member)
                # skip directories
                if not filename:
                    continue

                print "File: ", filename
                # copy file (taken from zipfile's extract)
                source = zip_file.open(member)
                with tempfile.NamedTemporaryFile(delete=False) as temp:
                    print temp.name
                    shutil.copyfileobj(source, temp)
                    tar.add(temp.name, arcname=filename)


gather_and_repackage_files("./stuff.zip", "./tarfile.tar")

Before I run this, the contents of my directory are "testin.py" (the program above) and "stuff.zip". "stuff.zip" is a zipfile containing two tiny text files, a.txt and b.txt, each of which contains about 15 characters. Apparently it also contains mac-backups of these, "_a.txt" and "_b.txt" as well (although when I expand it with the Archive utility, those do not appear, even with "ls -al").

After execution (Python 2.7.10), there's an additional file "tarfile.tar"; when I open this with the Archive utility on my Mac, I see this:

drwx------  6 jfh  staff  204 Oct 29 16:51 .
drwxr-xr-x  7 jfh  staff  238 Oct 29 16:51 ..
-rw-------  1 jfh  staff    0 Oct 29 16:50 ._a.txt
-rw-------  1 jfh  staff    0 Oct 29 16:50 ._b.txt
-rw-------  1 jfh  staff    0 Oct 29 16:50 a.txt
-rw-------  1 jfh  staff    0 Oct 29 16:50 b.txt

The temporary files created during execution actually DO contain the 15 or so characters of silly text, but the ones in the tarfile are zero-length.

So my question is "Why does the tar-file contain 0-length versions of a.txt and b.txt?"

Answer 1

The temp file may not have been completely flushed.

You could try to: temp.flush() os.fsync()

But of course it would be better not to create the temp file in the first place. Which you can avoid by using tar.addfile instead of tar.add .

You also need to set the size of the tarinfo that you provide.

Note: you could also set mtime to preserve the time.

This modification should do it:

import zipfile
import tarfile
import os

def gather_and_repackage_files(zip_file_path, target_file_path) :
    with tarfile.open(target_file_path, "w") as tar:
        with zipfile.ZipFile(zip_file_path) as zip_file:
            for info in zip_file.infolist():
                filename = os.path.basename(info.filename)
                # skip directories
                if not filename:
                    continue

                # copy file (taken from zipfile's extract)
                with zip_file.open(info) as source:
                  tarinfo = tarfile.TarInfo(filename)
                  tarinfo.size = info.file_size
                  tar.addfile(tarinfo, source)


gather_and_repackage_files("./stuff.zip", "./tarfile.tar")

Answer 2

Here is working code:

import zipfile, tempfile, shutil, tarfile, os

def gather_and_repackage_files(zip_file_path, target_file_path) :
    with tarfile.open(target_file_path, "w") as tar:
        with zipfile.ZipFile(zip_file_path) as zip_file:
            for member in zip_file.namelist():
                filename = os.path.basename(member)
                # skip directories
                if not filename:
                    continue

                print "File: ", filename
                print "Member: ", member
                source = zip_file.open(member)
                with tempfile.NamedTemporaryFile(delete=False) as temp:
                    print temp.name

                    shutil.copyfileobj(source, temp)

                    temp.close()
                    tar.add(temp.name, arcname=filename)

The secret sauce is in 'temp.close()', one line before the end. It turns out that that you can't add an open file to a tar archive (although the documentation doesn't seem to mention that).