split string

my old program is using cut to get the date of the file such:
for FILE in ATK*
do
file_year=$(echo $FILE| cut -c 13-16)
file_month=$(echo $FILE|cut -c 13-18)
###
echo $file_month --->just as example!
done
where filename is such:
ATK_7B1_WIP_20041116160700.1.xml.dsc
ATK_7B1_WIP_20041116160700.2.xml.dsc
ATK_7B1_WIP_20041116160700.3.xml.dsc
ATK_7B1_WIP_20041116160700.4.xml.dsc

problem is how can i enhance it to all filename?
such:
AIT_7B1_WIP_20050101010101.dsc
ASECL_7B1_WIP_20041209151027.xml.dsc
ASEKR_7B1_WIP_20050101020202.dsc

so i cant use the ' for FILE in ATK* ' rite?
should replace with what?
all the filename will be the same unless the front name:
xxxx_7B1_WIP_20041116160700xxx
help!!!

Filter out only the appropirate string of your digits and then you can cut the different fields as usualy. Try this script:

:
while read F; do
DIGITS=$(echo $F |
awk ' { match($0,"WIP_[0-9]+")
print substr($0, RSTART+4, RLENGTH-4)
}
'
)
echo $DIGITS
done <<EOF
AIT_7B1_WIP_20050101010101.dsc
ASECL_7B1_WIP_20041209151027.xml.dsc
ASEKR_7B1_WIP_20050101020202.dsc
EOF

As yu can see, I have used your file names. The script gives you the time stamp only. (Copy and store the script, then chmod +x and run it)

Have fun.

my version using sed

#!/bin/sh

fname=abc_20050201113355.xx
fname=lskdfldlfgjhlabc_20050201113355.xx

###REM a very generic way
###REM eval `echo $fname|sed 's/.*_\(....\)\(..\)\(..\)\(..\)\(..\)\(..\).*/Y=\1 M=\2 D=\3 h=\4 m=\5 s=\6/'`

###REM this is better, nota: sed does not know the '\d+' notation

eval `echo $fname|sed 's/.*_\([1-9][0-9]\{3\}\)\([0-9]\{2\}\)\([0-9]\{2\}\)\([0-9]\{2\}\)\([0-9]\{2\}\)\([0-9]\{2\}\).*/Y=\1 M=\2 D=\3 h=\4 m=\5 s=\6/'`

echo year $Y
echo month $M
echo day $D
echo hour $h
echo min $m
echo sec $s

Very clever, handy, faster than awk! But badly readable by a man.

To use awk or sed? It is a matter of taste and of personal skil. The computer time factor today is not too significant, because the saved seconds of computer time are cheaper than hours of programing.

If the program in the future should be administered by another person, I would prefer the awk version because of its better readability.

Regards

hello zlutovsky
i assume a programmer at least should know regexp
in *nix that's basics
so the long sed expr is really easy to read
awk is a formidable tool, in this case awk has no chance vs sed
because of the way awk is working
awk ... imagine an excel, split and full rows and fields, then work
sed ... read a line, scans cmds and apply eachh cmds on the readed line
as long you DONT need formatting or compilation, sed will be faster
i mean
a+=b; no probls in awk
you sure can convince 'sed' to do it, it's a longer way

i already solve my problem