topic Re: need some script help please in Operating System - HP-UX

need some script help please

Kristopher March — Mon, 17 May 2004 14:17:51 GMT

How can I place the value of two variables into one file on the same line?

See below:

I am trying to print two values to one line in a file. After the values are read into the file, I will cat the file and read the new values.

% cat citylist.txt

Columbus, OH 1999
Travis, TX 1998
Methuen, MA 2001
Salem, NH 2004

CITY=`cat citylist | awk '{print $1}'
CDATE=`cat citylist.txt | awk '{print $3}'
TWO=2
export CITY CDATE TWO
# I want to subtract field 2 ($2) by 2 years and then place the new value in a new file.

((NEWDATE=$CDATE-$TWO))

printf "%s %s \n" $CITY $NEWDATE > /tmp/newdate

cat /tmp/newdate

Columbus, OH 1997
Travis, TX 1996
Methuen, MA 1999
Salem, NH 2002

But this doesn't work. Where am I going wrong?

Re: need some script help please

D Block 2 — Mon, 17 May 2004 14:25:11 GMT

I could be wrong, but your first line in your INPUT file appears to be a blank-line.

your script seems fine to me, but when you print the NEWDATE as %s, you might want to treat it as a %d.

Re: need some script help please

Siddhartha M — Mon, 17 May 2004 14:29:02 GMT

Hi Kristopher,

If I understand your problem correctly you
are missing the "," separator. The citylist.txt has 2 fields not 3.
Thus when you do a

cat citylist.txt | awk '{ print $3 }'
it will throw a blank space.

You need to modify your code to take into
account the field separator ",".

Replace $3 with $2 in your code and check.

-Siddhartha

Re: need some script help please

John Poff — Mon, 17 May 2004 14:38:32 GMT

Hi,

Your command to get the city and state into the CITY variable actually will just grab the first word of the city. Using $2 will pick up the state, but only if your city names are all one word. Plus, you might want to put it into a loop so that you can work on the values of each line as you read them.

You could do something like this:

while read line
do
CITY=`echo $line | awk '{print substr($0, 1, length($0)-4}'`
CDATE=`echo $line | awk '{print $NF}'

((NEWDATE=$CDATE-$TWO))

printf ...(etc.)
done

Save that as a shell script and run it with the citylist.txt as input.

It's not pretty but it might do what you are looking for.

JP

Re: need some script help please

curt larson_1 — Mon, 17 May 2004 14:48:57 GMT

why don't you just do your calculation in awk. it will be easier and faster

cat citylist.txt | awk '
/NF > 2/ {$3=$3-2;print $0;
}' > /tmp/newdate

you might want to add some error checking to see if $3 is actually a number.

Re: need some script help please

Kristopher March — Mon, 17 May 2004 14:50:39 GMT

John - this is for you. Here's what I've got:

#!/bin/sh
#set -x
# name=testscript
#
while read line
do
CITY=`echo $line | awk '{print substr($0, 1, length($0)-4}'`
CDATE=`echo $line | awk '{print $NF}'`
TWO=2
export CITY CDATE TWO
((NEWDATE=$CDATE-$TWO))
printf '%s %s \n' $CITY $NEWDATE > newdate.txt
done

When I kick if off, it just sits at the prompt. Any other ideas?

Curt - I'll look into your suggestion now.

Points to come ...

Re: need some script help please

Kristopher March — Mon, 17 May 2004 15:02:05 GMT

Siddhartha - I can awk $3 of citylist.txt and get the year back. So I think that part is working.

Tom - what is %d as opposed to %s?

Curt - your idea seems to do something and then outputs an empty newdate file.

Re: need some script help please

curt larson_1 — Mon, 17 May 2004 15:04:19 GMT

if your going to do it in the shell, this would be better

TWO=2
while read CITY CDATE
do
((NEWDATE=$CDATE-$TWO))
printf '%s %d \n' $CITY $NEWDATE
done < citylist.txt > /tmp/newdate.txt

Re: need some script help please

curt larson_1 — Mon, 17 May 2004 15:06:21 GMT

my awk script might be off by 1 field, try

cat citylist.txt | awk '
/NF > 1/ {$2=$2-2;print $0;
}' > /tmp/newdate

Re: need some script help please

D Block 2 — Mon, 17 May 2004 15:08:00 GMT

The %d stays for 'd'ecimal value. the %s as you know, stays for 's'tring value.

str="abc"

printf "%s", str

foo=123456

printf "%d", foo

Re: need some script help please

John Poff — Mon, 17 May 2004 15:08:57 GMT

Kristopher,

If you named your script 'testscript', try it like this:

./testscript
JP

Re: need some script help please

Kristopher March — Mon, 17 May 2004 15:17:16 GMT

Curt - have a look:

CDATE=`cat citylist | awk '{print $3}'`
CITY=`cat citylist | awk '{print $1}'`
TWO=2
while read CITY CDATE
do
((NEWDATE=$CDATE-$TWO))
printf '%s %d \n' $CITY $NEWDATE
done < citylist > newdate.txt

which give me this:

%cat newdate.txt
Columbus,OH 0
Travis,TX 0
Methuen,MA 0
Salem,NH 0

Re: need some script help please

curt larson_1 — Mon, 17 May 2004 15:21:28 GMT

instead of
((NEWDATE=$CDATE-$TWO))

do

NEWDATE=$(($CDATE - $TWO))

Re: need some script help please

Kristopher March — Mon, 17 May 2004 15:22:05 GMT

I changed citlist to look only have two fields. plymouth,ma 2004

And the awk statement to read $2 instead of $3. But still same problem.

Re: need some script help please

Kristopher March — Mon, 17 May 2004 15:27:44 GMT

updated script:

CDATE=`cat citylist | awk '{print $2}'`
CITY=`cat citylist | awk '{print $1}'`
TWO=2
while read CITY CDATE
do
NEWDATE=$(($CDATE - $TWO))
printf '%s %d \n' $CITY $NEWDATE
done < citylist > newdate.txt

now I get:
./testscript2: syntax error at line 10: `NEWDATE=$' unexpected

Re: need some script help please

curt larson_1 — Mon, 17 May 2004 15:28:58 GMT

try this:

cat citylist.txt | awk '{print $2;}'

you should be getting the year

they try:

cat citylist.txt | awk '{a=$2-2;print $a;}'

then:

cat citylist.txt | awk '{$2=$2-2;print $2;}'

then

cat citylist.txt | awk '{$2=$2-2;print $0;}'

Re: need some script help please

curt larson_1 — Mon, 17 May 2004 15:32:07 GMT

you have a typo somewhere

you can comment these lines out as they are unnecessary

CDATE=`cat citylist | awk '{print $2}'`
CITY=`cat citylist | awk '{print $1}'`

to

#CDATE=`cat citylist | awk '{print $2}'`
#CITY=`cat citylist | awk '{print $1}'`

Re: need some script help please

John Poff — Mon, 17 May 2004 15:32:59 GMT

Curt has you on the right track. Try this:

awk '{$NF=$NF-2; print $0;}' citylist.txt

If that looks good, you can redirect the output to the /tmp/newdate file like this:

awk '{$NF=$NF-2; print $0;}' citylist.txt >/tmp/newdate

JP

Re: need some script help please

Siddhartha M — Mon, 17 May 2004 15:33:08 GMT

Check this:(seems to produce the desired output)

$cat citylist
Columbus,OH 1999
Travis,TX 1998
Methuen,MA 2001
Salem,NH 2004
$
$cat newdate.txt
cat: Cannot open newdate.txt: No such file or directory
$
$cat itrc
#!/bin/sh
CDATE=`cat citylist | awk '{print $2}'`
CITY=`cat citylist | awk '{print $1}'`
TWO=2
while read CITY CDATE
do
((NEWDATE=$CDATE-$TWO))
printf '%s %d \n' $CITY $NEWDATE
done < citylist > newdate.txt
$
$
$sh itrc
$
$cat newdate.txt
Columbus,OH 1997
Travis,TX 1996
Methuen,MA 1999
Salem,NH 2002

Re: need some script help please

John Poff — Mon, 17 May 2004 15:42:14 GMT

The trick with using $2 and $3 is ok if you only have city names that are one word. But what happens if your file includes?:

New Orleans, LA 1999

The NF variable in awk returns the number of fields read on a line, and the $NF variable is the value of the last variable on the line. If you use $NF in Curt's awk example, you can be sure that you will always get the year, no matter what the city name is.

Specifying the input file on the command line with awk saves using cat and piping it to awk.

JP