Shell Programming and Scripting

Hi,

I got the following problem.

I got a folder called "renapple" whose content has multiple ".sh" files in it:

they all contain one line of text: (this is just an example, the content itself looks different).

Code:

cp path1/file path2

What I want to do is:
- copy the content of all those files into new .sh file.
However every previous sh file should get its own newlin.

so the final sh should look a little bit something like this

Code:

cp path1/file path2
cp path1/file path2
cp path1/file path2
cp path1/file path2

Thanks in advance for any pointers.

Like this?

Code:

cat *.sh > final.sh

Thats a simple but effective solution. Thanks.

Will this also work with characters such as ä ü ä ß etc. ?

Unless you are guaranteed that final.sh will sort 1st in the list of files selected by *.sh (after final.sh is created), this cat will either duplicate some lines in final.sh or go into a loop repeatedly copying some lines from the start of final.sh into final.sh until the filesystem runs out of space or you hit the maximum file size limit for final.sh.

To perform this cat safely, the output file has to have a name that won't be matched by the pattern that is being given as an operand to cat. This can be done several ways; two include by directing the output to a file in another directory, as in:

Code:

cat *.sh > ../final.sh

or by using a different filename suffix for the target, as in:

Code:

cat *.sh > final$$ && mv final$$ final.sh

Any of these will work with any filenames ending in .sh.

KSH and Bash expansions are guaranteed to create an alphabetically sorted list of names before "feeding" it to cat.

Besides, file name expansion is performed before any redirection occurs.

Not sure about other shells.

Hi verdepollo,
You're correct. A POSIX conforming shell will perform pathname expansions before performing redirections.

Nonetheless, I have seen enough scripts fail because an output file was matched by an input pathname pattern expansion that I advise script writers to avoid the issue. When testing a new script and when a script fails for some reason and needs to be restarted, users frequently forget that they need to remove the output file before restarting the script. Even when it works, a user may want to run the script again the next day, week, or month and the output file can then accidentally become both an input file and an output file, leading to unexpected results. And if it is running on a system being shared with other users, the mistake can affect everyone using that filesystem; not just the user that forgot to remove the output file before restarting the script. Furthermore, when the script completes successfully, users often want to remove the (no longer needed) input files. If the output file matches the pattern that was used to select the input files, it is too easy to accidentally remove not only the input files, but also the output file.

Therefore, I strongly urge programmers not to create an output pathname that will be matched by the expansion of an input file pattern even though the output file should not be present when the script starts.

True, but we cannot blame the shell or a tool like cat for such behavior.

Personally, I wouldn't call it "safer use of cat" but poor script validation.

Yes.