Creating a self-extracting script to embed files in postings

Guenther Brunthaler · Apprentice Joined: 22 Jul 2006 Posts: 217 Location: Vienna

Hi all,

I have created a script which can used as a filter: It takes the output of tar as its input and writes a script as its output which, when run, will extract the files from the tar archive.

The scripts generated that way are very small and can be used as preformatted text in any posting, such as this one. Or they can be pasted directly into an e-mail message (which might be useful if you don't want to use attachments for any reason or if you are using a very dumb email program which does not support attachments).

My script does not assume anything about the contents of the tar archive passed to it by the user, i. e. it can contain text files, or binary files, or both.

The script will compress the archive contents passed to it with BZIP2 and GZIP (both with maximum compression enabled) and compare the compressed sizes against each other (and also against the uncompressed archive). It will then use the shortest of the resulting output versions for embedding into the script.

Which means that adaptive compression will be applied: The shortest resulting file size will be used, whether that means the tar archive has to be gzip-compressed, bzip2-compressed or not be compressed at all. (The latter can be true for archives already containing compressed files and which also happen to have a size which is close to being aligned to the tar output block size.)

The compressed archive will then be encoded as BASE-64 using OpenSSL, and finally a wrapper script will be added which will extract the files from the archive when run.

As a working example of the output of my script, here is a tar archive containing my script, after having been applied to itself:

Guenther Brunthaler · Apprentice Joined: 22 Jul 2006 Posts: 217 Location: Vienna

There is another tool called SHAR (for "SHell ARchive") which can also be used to convert any list of files into a self-extracting script file.

However, my script generates only 5 lines of overhead, where SHAR creates... MANY lines of overhead.

My script however assumes that the tar, gzip, bzip2 and openssl executables are available, or extraction will not work.

SHAR on the contrary, will work without any of those executables installed; it only requires a working sh.

This makes SHAR a better choice for barebone machines with only a minimum set of tools installed, but a workstation machine will quite certainly have all the tools installed which are required by my script.

And because my script has only very low overhead, it will typically be suited better for inclusion of a single file (or a set of rather small files) in postings.

Guenther Brunthaler · Apprentice Joined: 22 Jul 2006 Posts: 217 Location: Vienna

One might ask "why does your script not support compression using the compress tool, as used by tar -cZ?"

The answer is: Initially, I wanted to do that. But then I found out that compress was not installed on my box - although I have an otherwise pretty complete Gentoo installation.

In other words: While compress certainly is a standard tool in non-GNU UNIX systems, it seems not to be installed on every GNU/Linux system.

And that's why I chose not to support it in my script either.

However, it can easily be added by applying the following patch:

truc · Advocate Joined: 25 Jul 2005 Posts: 3199

you already used this trick on the forums before, and I liked it. this can be really handy!

thank you
_________________
The End of the Internet!

avx · Advocate Joined: 21 Jun 2004 Posts: 2152

Seen it - tried it - loved it ... do I need to say more?
Well, a big thank you comes to my mind, so here it is: Thank you

bludger · Guru Joined: 09 Apr 2003 Posts: 389

This looks pretty cool.

When I first tried pasting this into a file and extracting it, I got the following error:

gzip: stdin: unexpected end of file
bbs-package-tar
tar: Child returned status 1
tar: Error exit delayed from previous errors

I then saw that there was a space or something after the "END". When I deleted this, it extracted without any error messages. Is there some way of getting around this problem?

Guenther Brunthaler · Apprentice Joined: 22 Jul 2006 Posts: 217 Location: Vienna

Guenther Brunthaler · Apprentice Joined: 22 Jul 2006 Posts: 217 Location: Vienna

Hi all,

I found a working solution to the whitespace-problem: Just remove the "END" line from the last line of the generated script, and it will work no matter how much whitespace is added by the copy/paste to the end of each line!

As it seems, the "here doc"-syntax works without the delimiter as well: In this case, all the data until EOF will be read (and fed to OpenSSL); which is exactly what I want.

However, I do not know whether this behaviour is portable, or just coincidental.

An ideas about the portability?

If it is indeed portable, I will happily modify my script to omit the ending delimiter for the "here doc" text. This will even make the generated script one line shorter - with no additional changes required.

UPDATE:

In http://opengroup.org/onlinepubs/007908799/xcu/chap2.html is stated that a missing delimiter is implementation-defined and should be considered an error.

Which means, this approach to the whitespace-problem just isn't going to be the right one. ;-(