Shell Programming and Scripting

I think I am clear with your issue. You refuse to see that the backslash character (\) and the slash character (/) are different.

You say that the following:

Code:

mv ${DATA_DIR}\ABC_*.dat ${ARCHIVE}

works on the command line, but don't understand why the following:

Code:

mv ${DATA_DIR}/ABC_*.dat ${ARCHIVE}

doesn't work in your script.

Please execute the following commands exactly and post the results:

Code:

uname -a
ls -l /u01/data/oracle/rms/UPGRMSBATCH/RETLforRPAS/data\ABC_*.dat
ls -l /u01/data/oracle/rms/UPGRMSBATCH/RETLforRPAS/dataABC_*.dat
ls -l /u01/data/oracle/rms/UPGRMSBATCH/RETLforRPAS/data/ABC_*.dat

On UNIX and Linux systems, the / character separates components in pathnames. On UNIX and Linux systems, the shell uses \ as an escape character; not to separate a directory name from the name of a file in that directory. Therefore, the command:

Code:

mv ${DATA_DIR}\ABC_*.dat ${ARCHIVE}

and the command:

Code:

mv ${DATA_DIR}/ABC_*.dat ${ARCHIVE}

are VERY different. If the first one is working on the command line and the second one is not working in your script; use the first one in your script.
Note that the first one uses the asterisk (*) to expand to a list of files that match the pattern dataABC_*.dat in the directory:

Code:

/u01/data/oracle/rms/UPGRMSBATCH/RETLforRPAS

while the second one uses it expand to a list of files that match the pattern ABC_*.dat in the directory:

Code:

/u01/data/oracle/rms/UPGRMSBATCH/RETLforRPAS/data

Hi All,

Thanks for you help. Here is the answer,

Code:

ls ${DATA_DIR} | grep "ABC" | while read x; do echo $x; mv ${DATA_DIR}/$x ${ARCHIVE}; done

Thanks,
Bopty

If the above command works, your current working directory has to be $DATA_DIR. In that case, the commands:

Code:

printf "%s\n" $PWD/*ABC*;mv *ABC* $ARCHIVE

will do the same thing MUCH more efficiently, but it will print all of the filenames before it moves any of them.

If it is important to list each filename just before it is moved, the MUCH less efficient commands:

Code:

for i in *ABC*
do      echo $PWD/$i
        mv $i $ARCHIVE
done

will still be more efficient than the script you're currently using.

And if there is not any, test it or you get some unwanted action. Example:

Code:

for i in *ABC*
do     
        [ "$i" = "*ABC*" ] && continue  # there wasn't any *ABC* files 
        echo $i
        mv $i $ARCHIVE
done

If file name generation result is 0 files, then what you are doing ?
You are doing something for filename *ABC*.

Code:

touch *ABC*   # change files timestamp if you have *ABC* files
                    # but if not, touch will create filename *ABC*

Good point. From the description of the problem, I was under the impression that there would always be at least one match, but it would be better to verify that there is a match. The test can be inside the loop, or an if could be added before the loop that exits if there is no match. I believe some shells also have a set special built-in utility option that causes patterns that contain wildcard characters to be removed from the list if there are no matching files.

In my script, if there is no match, $PWD/*ABC* will be printed (with the asterisks) and mv will print a file not found diagnostic message and terminate with a non-zero exit status causing the script to return a non-zero exit status.