Re: How to convert files

[Bob McGowan]

FuziOK wrote:

John Hasler wrote:

FuziOK writes:

I'm not clear how to use command file to detect binaries.

Use it to detect DOS files.  From the man page:

    If the lines of a file are terminated by CR, CRLF, or NEL, instead of
    the Unix-standard LF, this will be reported.

Or parse file's output for the identifier for your binaries, if they are of
predictable types.

What is the problem you are trying to solve?
--
John Hasler

Thanks for your response.

In fact, I am puzzled with binary files and text files:
How can I say a file is a binary file, not a text file?

Example:

$ (cat /bin/ls; echo '#!/bin/sh') > aaa
$ file aaa

$ flip -m aaa
aaa: binary file, not converted

$ (echo '#!/bin/sh'; cat /bin/ls) > bbb
$ file bbb
bbb: Bourne shell script text executable
$ flip -m bbb
bbb: binary file, not converted

aaa: ELF 32-bit LSB executable, Intel 80386, version 1 (SYSV), for GNU/Linux 2.4.1, dynamically linked (uses shared libs), for GNU/Linux 2.4.1, stripped From the example, I think, at least, 'file' is not suitable tool to dectect binary files.

So, you did a 'conversion' on the binary file, making it a mixed text/binary. I think this would need to be thought of as a 'special case', since mixed text/binary files are not too common (but see last paragraph). The 'file' command will look at the initial bytes in a file, until it reaches a point where it has a match with some string in it 'magic strings' data base (/usr/share/file/magic.mgc and /etc/magic), and reports that. It doesn't try to be overly smart, it wants to be fast. And generally, it's quite accurate. On the other hand, 'flip' has to process the whole file, and if any part is binary, conversion of the whole would be a problem. So it reads the whole file first, I suspect, rather than just searching for a lead in 'magic' string value. Given the original requirements, I'd agree that using 'file' would be a problem, so I'd depend on 'flip' for determining the text/binary question. FYI, I do know of mixed file types, sometimes used to provide binary code with a preamble shell script to extract it. Nvidia does this with their proprietary video drivers. I have one that has 747 lines of sh script, and 'flip' still reports it as 'binary', so I'm guessing (having not examined the code) that it reads the whole file before making any conversion attempt. This is the safe thing to do, since you *don't* want changes made in the binary part, under any circumstances. This is not the best thing to do for general identification of a file's content, such as the 'file' command does.

smime.p7s
Description: S/MIME Cryptographic Signature