UNIX for Dummies Questions & Answers

I have a directory

Code:

/mo/omar/teefa/filesDBM

it contains files that all start with BMW and ends with .txt for example

Code:

IBM0001.txt
IBM0002.txt
...............

my script test.sh

Code:

file=$1       # source file 
file=$2       # temp file will be removed 
file2=$3     # output file 

awk ' /testname/ { print $1 } /TypeModel/ { print ":" $3 } ' / $file >> $file1
paste -d ";" $file $file1 >> $file2
rm $file1

i wanna make the output file to be named automatically by date for example

Code:

22112011.txt             or 12011.txt
23112011.txt             or 22011.txt

and i wanna make it runs automatically every day at 10 am
[ each day at 10 am ./test.sh ]
and to put its input automatically by making a script to search in the given directory for the files starts with BMW and ends with .txt [ take care make it to rename files . processing and .done ] to prevent that file it taken twice

finally: a scripts search certain source files and runs every day at 10 am and all ouput in a file named by date thanks alot waiting your brilliant replies

Try looking in to the cron command to schedule your script to run.

Hi you can use the date command inside backticks (`) in order to provide a date as filename for the output file, i.e.:

Code:

/path/to/test.sh 1st_param 2nd_param `date '+%Y%M%d'`.txt

resulting in the third parameter being expanded by the shell as: 20111128.txt (that is YYYYMMDD.txt).

See man pages for date command for formatting date output.
see ya
fra

thanks alot
what if i wanna get all ,txt files in this directort without overwritting

hi,

i don;t know if I fully understood your last question, but you could think about using hour, date and time in the output files in order to differentiate the filenames among different executions of the script in the same day.

fra

ok vor this part for output
i want my input files to be taken from a certain path ( all starts with IBM and end .txt )
and take them by order without overwriting (taken twice ) . is it ok now ?

ok,
so you could just use a for cicle to call your script, like:

Code:

for sourcefile in `ls /mo/omar/teefa/filesDBM/IBM*.txt` ; do /path/to/test.sh $sourcefile 2nd_param 3rd_param ; done

The shell will expand the

Code:

`ls /mo/omar/teefa/filesDBM/IBM*.txt`

by returning all the filenames contained in the path.
Be aware: the ls commend will return both files and subdirectories as well as links contained in /mo/omar/teefa/filesDBM, so if you want only filenames returned, use the find command instead (see man find with particular reference to the -t, -name and -print options).

Obviously, you can use the `date` trick to obtain output filenames based on dates.
You can use the previous one-liner right in the crontab.

see ya
fra

P.S.
please be also aware of the -maxdepth option in find (it limits the scope to the directory passed as argument, thereby avoiding subdirectory recursion).
Finally, ls without other time-related switches guarantees that filenames are returned in alphabetical order (by the way, digits are actually considered strings); I'm not sure that the -name switch of find does the same.