Compare directories the most efficient way

Compare directories the most efficient way

[Meino Christian Cramer]

Hi,

 (using Linux)

 I have a problem and dont know, whether zsh includes some "extras" to
 solve it.

 There are two identical directory trees with binary stuff in it but
 with a different prefix for example:
 
  /home/user/onetree/. 

 and 

  /home/user/theothertree/.

 Unfortunately the files in there have "invalid" filenames with spaces
 and other ASCII- and not-so-ASCII-traps (german Umlauts for example).

 It is not possible to rename the files with detox, since then other
 things will no longer work.

 I want to binary-compare all files of one tree with those
 corresponding ones of the other tree. For this purpose I choose the
 "cmp" command (but if there are others more zshy ones...I have no
 problem to use those.... :O).

 The problem is to write a function/routine/script or whatever to do
 that job. Everything tried result in errors like:

   This   : file not found 
   is     : file not found 
   one    : file not found
   file   : file not found 
   with   : file not found 
   spaces : file not found 
   .bz2   : file not found


 I tried "find <path> -ls | cut -b 69- " as a source for escaped filenames
 in a for/do/done-loop but it failed the same way as mentioned above.

 Is there any way out? :)

 Thank you very much for any help in advance.

 Keep zshing!
 Meino

 
 

Re: Compare directories the most efficient way

[Jean Chalard]

>  There are two identical directory trees with binary stuff in it but
>  with a different prefix for example:
>   /home/user/onetree/.
>  and
>   /home/user/theothertree/.

I think I may have missed something here, because I can't see why something like

for i in onetree/**/*; do cmp -- "$i" "${i/onetree/theothertree}"; done

...wouldn't work.
It wouldn't split the filenames on spaces, and I don't think any ascii
or non-ascii bizarreness would affect that kind of thing.


>  The problem is to write a function/routine/script or whatever to do
>  that job. Everything tried result in errors like:
>
>    This   : file not found
>    is     : file not found
>    one    : file not found
>    file   : file not found
>    with   : file not found
>    spaces : file not found
>    .bz2   : file not found
>
>  I tried "find <path> -ls | cut -b 69- " as a source for escaped filenames
>  in a for/do/done-loop but it failed the same way as mentioned above.

As I see it this is most likely caused by the sh_word_split option set
(which you probably *really* don't want most of the time), and the
absence of any quoting mark.

--
J
"Toi, je te trouve pas la même tête que sur la page précédente" -- Wakamiya

Re: Compare directories the most efficient way

[Vincent Lefevre]

On 2005-10-25 12:27:28 +0900, Jean Chalard wrote:
> >  There are two identical directory trees with binary stuff in it but
> >  with a different prefix for example:
> >   /home/user/onetree/.
> >  and
> >   /home/user/theothertree/.
> 
> I think I may have missed something here, because I can't see why
> something like
> 
> for i in onetree/**/*; do cmp -- "$i" "${i/onetree/theothertree}"; done
> 
> ...wouldn't work.

Several problems:
  * If there's a file in theothertree that doesn't exist in onetree,
    you won't see the corresponding error.
  * I think you should use something like onetree/**/*(D^/) to include
    files starting with a dot, but not directories.
  * If there are special files (symlinks...), this is even more
    complicated.

"diff -r" (possibly with -q) from the diffutils is probably a better
choice.

-- 
Vincent Lefèvre <vincent@vinc17.org> - Web: <http://www.vinc17.org/>
100% accessible validated (X)HTML - Blog: <http://www.vinc17.org/blog/>
Work: CR INRIA - computer arithmetic / SPACES project at LORIA

Re: Compare directories the most efficient way

[Meino Christian Cramer]

From: Vincent Lefevre <vincent@vinc17.org>
Subject: Re: Compare directories the most efficient way
Date: Tue, 25 Oct 2005 10:45:27 +0200

> On 2005-10-25 12:27:28 +0900, Jean Chalard wrote:
> > >  There are two identical directory trees with binary stuff in it but
> > >  with a different prefix for example:
> > >   /home/user/onetree/.
> > >  and
> > >   /home/user/theothertree/.
> > 
> > I think I may have missed something here, because I can't see why
> > something like
> > 
> > for i in onetree/**/*; do cmp -- "$i" "${i/onetree/theothertree}"; done
> > 
> > ...wouldn't work.
> 
> Several problems:
>   * If there's a file in theothertree that doesn't exist in onetree,
>     you won't see the corresponding error.
>   * I think you should use something like onetree/**/*(D^/) to include
>     files starting with a dot, but not directories.
>   * If there are special files (symlinks...), this is even more
>     complicated.
> 
> "diff -r" (possibly with -q) from the diffutils is probably a better
> choice.

Hi,

 yes, indeed....there are dotted files in there...even directories are
 dotted...and not found with the pattern above.

 I didn't choose diff, since I wanted a more common (or is "general"
 the better choice?)  solution in case of haveing the task to feed two
 corresponding files to a program, which "does" something with that
 input.

 keep zshing!
 Meino


 
> -- 
> Vincent Lefèvre <vincent@vinc17.org> - Web: <http://www.vinc17.org/>
> 100% accessible validated (X)HTML - Blog: <http://www.vinc17.org/blog/>
> Work: CR INRIA - computer arithmetic / SPACES project at LORIA
>