Query on the file

[Chandrakumar Karunamoor]

Hi all,
I have a comma delimited file in which the first field is a key field. I need to do something equivalent to the following in unix.

select count(*) from where is between and ;

The start and end values could be anything (except that end value is always greater). I need to split the file depending on the number i get from above.

In addition to the count if I get the line numbers of the records for the start and end that would be great.

Something worth mentioning here is the start and end value may or may not be in the file.

For example if below is the file

201,34934....
204,39438...
207,983493...
210,03943...
220,0983.....

If my start and end values are 202 and 215. The result needs to be 3 for the count and 2 and 4 for the line numbers.

Thank you very much for you help.
Yours Sincerely,
Chandra

[harry d brown jr · Accepted solution]

How about something as ugly as this:

#!/opt/perl/bin/perl
#
$begVAL=$ARGV[0];
$endVAL=$ARGV[1];
$filename=$ARGV[2];
open(fileptr,"<$filename");
#
$begno=0;
$endno=0;
$lineno=0;
#
while () {
$lineno += 1;
($fld1val, $fldrst) = split(/,/);
if (( $begVAL <= $fld1val ) && ( $begno == 0 )) {
$begno = $lineno;
}
if ( $endVAL >= $fld1val ) { $endno=$lineno; }
}
print "first is $begno\n";
print "last is $endno\n";
print "num lines is ",$endno-$begno+1,"\n";

./someperl.pl 202 215 ./numlist.txt
first is 2
last is 4
num lines is 3
#


live free or die
harry

[A. Clay Stephenson]

I would probably use awk for this:

Create an awk script, my.awk

BEGIN {
tot = 0
start = -1
end = -1
}
{
val = $1 + 0
if ((val >= lo) && (val <= hi))
{
++val
if ((start < 0) && (val == lo)) start = NR
if (val == hi) end = NR
}
}
END {
printf("%d lines found ",tot)
if (start >= 0) printf("Starting line %d ",start)
if (end >= 0) printf("Last key found %d",end)
printf("\n")
}

Then invoke awk thusly:

awk -F, -f my.awk -v "lo=202" -v "hi=215" myinputfile

[Rodney Hills]

"perl" does not have the 18 digit limit. If you can rewrite your script in "perl" you'll have other tools at your disposal too.

-- Rod Hills

[Leif Halvarsson_2]

Hi,
Or a shell script, call it "myscript" and do:
myscript



IFS=","
typeset -i cntr a
cntr=0

while read a b
do
if [[ $a -ge 202 && $a -le 215 ]]
then
cntr=$cntr+1
fi
done
echo $cntr

[Chandrakumar Karunamoor]

Thank you Harry and Clay.

I have a quick question in korn shell. I don't think korn shell supports integers of 18 digits. Is there a work around for this in korn shell itself.

Any help is greatly appreciated.

Yours sincerely,
Chandra

[A. Clay Stephenson]

There is no quick fix for that problem. The various shells support 32-bit signed integers whichs give you a range of ~ -2 billion .. +2 billion. If you need more than that then you can use bc for essentially unlimited precision and range.

[Leif Halvarsson_2]

Sorry,

I did not read your description carefully enough.
Here is an example shell script with start and end line numbers.


IFS=","
typeset -i cntr a start end line
cntr=0
start=0
end=0
line=0

while read a b
do
line=line+1
if [[ $a -ge 202 && $a -le 215 ]]
then
cntr=$cntr+1
if [ $start -eq 0 ]
then
start=$line
fi
end=$line
fi
done
echo Nr: $cntr Start: $start End: $end


No points for this .

Leif