Long file names within shell script

I am downloading a zip file that contain files that are very long. I am trying to process them, but cannot. I can move the files from one directory to another at the shell prompt, but not within a shell script, I get a stat error.

The files look somewhat like this;

08fe457ab_34fed678_MCRXCOMM.FIL_999.edi

I do not know what the problem is.

Any help would be appreciated!

---------- Post updated at 08:51 PM ---------- Previous update was at 07:59 PM ----------

I was able to resolve this problem.

I was trying to delete the original post, but did not know how.

Don't! Write your solution here instead, so that others, which have the same (or a similar) problem can learn from your findings.

bakunin

Not sure how I resolve my issue. I was trying many things and it just finally worked. I was using the following code;

mv $filxfr/*.edi  $filtmp/.
for RPTFIL in `ls`
do
    mv $RPTFIL FLXRPT.DAT
    rpt999
       .
       .
       .
done

In the filxfr directory, I had zip and text files. So the above code would not work. I made sure the text files were in another directory for processing. I later found an error in my subroutine (ie. rpt999) that was not looking in the correct directory.

Again, I am not sure this helps. I would have posted this last night, but I thought it would not be helpful. That is why I wanted to delete my original post. I am a firm believer in sharing results or ideas.

First off: thanks for sharing. And because you did that (instead of just saying you have a solution) you get something back immediately in return: some explanation why your script probably hasn't worked. So this will be at least helpful to you (but perhaps some others, because what you did is a common error). Here it goes:

When you write scripts, as a rule of thumb, NEVER use relative paths or filenames. See, you can call a script from everywhere and its workings should always be the same, regardless of from where you called it. The line

for RPTFIL in `ls`

will work if you call the script in one directory, but fail if you call it in another. Save for that, you should avoid the backticks (they are obsolete) but you do not need them anyway: for file in *edi ; do will work the same and you do not need to call ls for that. You see, "*" means already "all files" and is interpreted b the shell, not the program called. A command like:

ls *file

is processed this way: first, the shell sees the asterisk and "expands" it to a list of files matching, in this case: all files with names ending in "file" in the current directory. Then the result of this "expansion" is put on the (intermediate) command line:

ls *file                     # your command
ls a.file b.file c.file      # after expansion (you won't see that)

Finally the command itself (here ls ) is called and given the list of files, which it displays. If you call ls without any parameters it doesn't do much at all, only when you use additional options it gets more and more useful. But try, to test what i have said, on the commandline:

echo *

and you will see that it produces the same list as ls without options. And with the explanation above you know why.

Anyway, preferable is to provide what the script needs to know on the commandline, like:

myscript.sh /path/to/my/data

UNIX has a quite fixed directory hierarchy and there is a certain place for all sorts of things. You put i.e. temporary files always in /tmp and, on the other hand, whatever is in /tmp can be deleted without question. So your script can assume some paths to be given.

Second: as a rule of thumb, always quote your variables! The shell treats space as separators, so a line like:

a b c

will call a program "a" with the first parameter "b" and the second parameter "c". It is possible to name a program "a b" or pass parameters with enclosed spaces you need to quote:

"a b" "c d"    # will start a program "a b" with a single parameter "c d"

Now let us see your script itself:

The first thing is: when you start programming, keep a strict order. It will help you to organize your code. We always start with a shebang line, telling the OS which shell to use (instead of relying on an arbitrary default that could change), then declare the variables we use. This gives us the opprtunity to write in commentary what should go into these variables. Note that everything after "#" is a commentary.

#! /bin/ksh
# maybe you want to use "#! /bin/bash" instead, this line is called "shebang"

dir="$1"                     # working directory
tmpdir="/tmp/mydir"          # temporary workspace
file=""                      # filename buffer

if [ -e "$tmpdir" ] ; then
     mkdir -p "$tmpdir"
fi

mv ${dir}/*edi  "$tmpdir"

for file in "$tmpdir/*" ; do
    mv "file" "/some/place/FLXRPT.DAT"
    process_file
    if [ $? -ne 0 ] ; then         # you might want some error reporting here
        echo "ERROR processing $file" >2&
    fi
    ....
done

rmdir "$tmpdir"

exit 0

This will first copy all files "*edi" from some directory to "/tmp/mydir", then copy one file after other to the name "/some/place/FLXRPT.DAT" and process it. if the (hypothetical) command "process_file" exits with any other value than 0 (conventionally this means success) an error message is produced and the script continues (it might also stop if you put an "exit" command after the echo). Once the list is finished the temporary directory is deleted and the script exits with 0 itself.

Now, i suppose this is not exactly what you wanted to do and you could (and should) modify it to better suit your needs. It is not so much about how to write a certain script but to show the princinples of how to write any script. Always bear in mind that scriptingis like any other form of software engineering and the same principles apply. In fact it is even more demanding (mostly on self-control) because you can get away initially with a lot of slack where other programming languages are more demanding from the start. You should mostly not rely on these shortcuts, because the longer your scripts get (and, believe me, once you get the hang of it you are NEVER going to stop) the more you need to drop these bad habits and do it the correct way. So better start developing good habits from the start.

In either case, if you have troubles extending your script we will be here to help you.

I hope this helps.

bakunin

the shebang line needs to be corrected, do not place a space between exclamation mark and slash, type it like this

 
 #!/bin/ksh
 

Sorry, but this is not correct: in fact both shebang lines:

#! /path/to/interpreter

and

#!/path/to/interpreter

will work because the three-byte "magic number" "#!/" as well as the 4-byte magic number "#! /" is built into the UNIX-kernel. You need to use absolute pathes, though, and i.e.

#! ../path/to/interpreter

will not work, regardless of a space being there or not.

I hope this helps.

bakunin