duplicity-talk
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Duplicity-talk] gpg key password asked for backup after verify


From: edgar . soldin
Subject: Re: [Duplicity-talk] gpg key password asked for backup after verify
Date: Wed, 31 May 2017 17:32:11 +0200
User-agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:52.0) Gecko/20100101 Thunderbird/52.1.1

hey Ken,

i tested the following on a local cygwin installation and it patched away fine 
:)

'duplicity/librsync.py' line 159+
"
class PatchedFile(LikeFile):
    """File-like object which applies a librsync delta incrementally"""
    def __init__(self, basis_file, delta_file):
        """PatchedFile initializer - call with basis delta

        Here basis_file must be a true Python file, because we may
        need to seek() around in it a lot, and this is done in C.
        delta_file only needs read() and close() methods.

        """
        LikeFile.__init__(self, delta_file)
        if not isinstance(basis_file, types.FileType):
            if hasattr(basis_file, 'file') and isinstance(basis_file.file, 
types.FileType):
                basis_file = basis_file.file
            else:
                raise TypeError("basis_file must be a (true) file or an object 
whose file attribute is the underlying true file object")
        try:
            self.maker = _librsync.new_patchmaker(basis_file)
        except _librsync.librsyncError as e:
            raise librsyncError(str(e))
"

is that ok w/ you? do you need a branch?  ..ede

On 29.05.2017 16:47, Kenneth Loafman wrote:
> I don't know what the underlying code looks like for non-POSIX systems, and I 
> don't have a Windows/Cygwin system to test against, so my answer would be 
> that I would /prefer/ it be done in /_librsyncmodule.c/ since that seems to 
> be the only place that needs it.  Failing that, /librsync.py/ is my second 
> preference since it's only one level up.
> 
> We would still leave the error check  on line #305 since we could have made a 
> different error.  We would need to set the error message accordingly.
> 
> ...Ken
> 
> 
> On Mon, May 29, 2017 at 8:47 AM, edgar.soldin--- via Duplicity-talk 
> <address@hidden <mailto:address@hidden>> wrote:
> 
>     Ken,
> 
>     On 29.05.2017 14:27, edgar.soldin--- via Duplicity-talk wrote:
>     > On 29.05.2017 13:26, Kenneth Loafman wrote:
>     >> 1) This was from the README for 0.7.05:
>     >>
>     >>         * Fix bug #1494228 CygWin: TypeError: basis_file must be a 
> (true) file
>     >>
>     >>           - The problem that caused the change to 
> tempfile.TemporaryFile was due
>     >>
>     >>             to the fact that os.tmpfile always creates its file in the 
> system
>     >>
>     >>             temp directory, not in the directory specified.  The fix 
> applied was
>     >>
>     >>             to use os.tmpfile in cygwin/windows and 
> tempfile.TemporaryFile in all
>     >>
>     >>             the rest.  This means that cygwin is now broken with 
> respect to temp
>     >>
>     >>             file placement of this one file (deleted automatically on 
> close).
>     >
>     > wrt. os.tmpfile() .. afaics. does
>     >
>     >    https://hg.python.org/cpython/file/2.7/Lib/tempfile.py#l484 
> <https://hg.python.org/cpython/file/2.7/Lib/tempfile.py#l484>
>     >
>     > use the os module ( os.open, os.fdopen ) all around to create temp 
> files. aren't these supposed to return a (true) file? the docs say nothing in 
> this regard. https://docs.python.org/2/library/os.html#file-object-creation 
> <https://docs.python.org/2/library/os.html#file-object-creation>
>     >
>     > ..ede
>     >
> 
>     ok answering myself here, after reading the original bug report 
> thoroughly, which says
>     "
>     The problem is that, as explained in the Python documentation 
> (http://docs.python.org/library/tempfile.html 
> <http://docs.python.org/library/tempfile.html>), tempfile.TemporaryFile 
> returns a *file-like* object, not a file object. Under Linux, it effectively 
> returns a file object but under windows it does not. Hence, the type of the 
> object returned is *not* a types.FileType, causing the exception to be raised.
>     "
>       https://bugs.launchpad.net/duplicity/+bug/670891 
> <https://bugs.launchpad.net/duplicity/+bug/670891>
> 
>     further reading shows that the error stems from 'duplicity/librsync.py'
>     "
>       File "/usr/lib/python2.6/site-packages/duplicity/librsync.py", line 
> 167, in __init__
>         raise TypeError("basis_file must be a (true) file")
>     "
> 
> 
>     so obviously something is done differently in the 
> 'duplicity/_librsyncmodule.c', that needs a delta patch file to be a plain 
> File object. let's see 'duplicity/librsync.py' says
>     "
>     class PatchedFile(LikeFile):
>         """File-like object which applies a librsync delta incrementally"""
>         def __init__(self, basis_file, delta_file):
>             """PatchedFile initializer - call with basis delta
> 
>             Here basis_file must be a true Python file, because we may
>             need to seek() around in it a lot, and this is done in C.
>             delta_file only needs read() and close() methods.
> 
>             """
>             LikeFile.__init__(self, delta_file)
>             if not isinstance(basis_file, types.FileType):
>                 raise TypeError("basis_file must be a (true) file")
>             try:
>                 self.maker = _librsync.new_patchmaker(basis_file)
>             except _librsync.librsyncError as e:
>                 raise librsyncError(str(e))
>     "
>     
> http://bazaar.launchpad.net/~duplicity-team/duplicity/0.8-series/view/head:/duplicity/librsync.py#L163
>  
> <http://bazaar.launchpad.net/%7Eduplicity-team/duplicity/0.8-series/view/head:/duplicity/librsync.py#L163>
> 
> 
>     and in 'duplicity/_librsyncmodule.c' it tries to utilize 
> 'PyObject_AsFileDescriptor(python_file)'
>     "
>     /* Call with the basis file */
>     static PyObject*
>     _librsync_new_patchmaker(PyObject* self, PyObject* args)
>     {
>       _librsync_PatchMakerObject* pm;
>       PyObject *python_file;
>       int fd;
> 
>       if (!PyArg_ParseTuple(args, "O:new_patchmaker", &python_file))
>         return NULL;
>       fd = PyObject_AsFileDescriptor(python_file);
>     "
>     
> http://bazaar.launchpad.net/~duplicity-team/duplicity/0.8-series/view/head:/duplicity/_librsyncmodule.c#L294
>  
> <http://bazaar.launchpad.net/%7Eduplicity-team/duplicity/0.8-series/view/head:/duplicity/_librsyncmodule.c#L294>
> 
> 
>     so, if i understand correctly, we just have to implement the method 
> 'fileno()' as described here
>       https://docs.python.org/2/c-api/object.html#c.PyObject_AsFileDescriptor 
> <https://docs.python.org/2/c-api/object.html#c.PyObject_AsFileDescriptor>
>     in another wrapper around the tempfile.TemporaryFile() return value
>       https://docs.python.org/2/library/tempfile.html#tempfile.TemporaryFile 
> <https://docs.python.org/2/library/tempfile.html#tempfile.TemporaryFile>
>     , utilizing (if necessary) the file-like object's file attribute as 
> mentioned in the python doc and return
>     "
>     The returned object is a true file object on POSIX platforms. On other 
> platforms, it is a file-like object whose file attribute is the underlying 
> true file object.
>     "
> 
>     implementing that should render the failing TypeError("basis_file must be 
> a (true) file") in 'duplicity/librsync.py' meaningless, so it could be 
> removed or?
> 
>     alternatively, we could simply extend 'duplicity/librsync.py', to test 
> for a file attribute, in case the "file" object given is just a wrapper. on 
> success we then simply deliver that (pseudo code)
>     "
>     self.maker = _librsync.new_patchmaker(basis_file.file)
>     "
> 
>     what do you think Ken? would that work? which approach would you prefer? 
> ..ede
> 
>     _______________________________________________
>     Duplicity-talk mailing list
>     address@hidden <mailto:address@hidden>
>     https://lists.nongnu.org/mailman/listinfo/duplicity-talk 
> <https://lists.nongnu.org/mailman/listinfo/duplicity-talk>
> 
> 




reply via email to

[Prev in Thread] Current Thread [Next in Thread]