Shell Programming and Scripting

Dear All,
I have a log file like below

Code:

13:26:31 |152.22
13:27:31 |154.25
13:28:31 |154.78
13:29:31 |151.23
13:30:31 |145.63
13:31:31 |142.10
13:32:31 |145.45

where values will be there from 00:00 hrs to 23:59 hrs. I'm matching for last occurance of 23:59 and printing 1440 lines (grep -B1440 "23:59").. But if values are missing from 00:00 hrs to 2:00 hrs... previous day values also gets printed.

How to avoid this prob???

Well, you have an unreliable data set, so you need to analyze it without any speific time. You could use awk or shell to find each pair of lines when time folds back, an apparent day boundary. (Some like to log the date like "YYYY-MM-DD_HH:MM:SS ($$) ...." in every line, so this situation is prevented.) This finds the boundary lines:

Code:

lntm=0
while read tm pval
do
 xtm=${tm#*:}
 ntm=${tm%%:*}${xtm%:*}${xtm#*:}   # make time tm look like an integer ntm
 if (( lntm > ntm ))   # is last numeric time greater (fold back)?
 then
  echo $ltm $lpval
  echo ----
  echo $tm $pval
  echo . . .
 fi
 lntm=ntm ltm=tm lpval=pval
done <$in_file

Little more explanation will be helpful.... whether pval is a command?

pval is a shell variable, payload value. read puts values in variables, using the shell's usual $IFS characters as separators, and putting excess values, at least with ksh but it varies, in the last variable, so "read a b" is read first field into $a and the rest into $b. My personal mental nomenclature is that record fields are, at any given oment, either keys (needed to address this record) or payload (stuff that serves the application but not this support function, along for the ride. The key is part of the delivery mechanism, like the space shuttle, and the rest is cargo, goes in the bay to follow the key to the destination.


Now, an alphanumeric compare can compare times without removing the :, but I was lazy and had not gotten into researching string compare, a sin in the Internet Age, so now we both learn. It seems "[[ $a < $b ]]" works fine, especially with such bland strings. Depending on shell and maybe locale, '[[ "A" < "A" ]]' and '[[ "3" < "c" ]]' might not be what yoou think. The shell has several parallel comparison suites:So, the narrative flow is:Revised as noted for string compare:

Code:

ltm=0
while read tm pval
do
if [[ $ltm > $ntm ]]   # is last time greater (fold back)?
then
  echo $ltm $lpval
  echo ----
  echo $tm $pval
  echo . . .
fi
ltm=tm lpval=pval
done <$in_file

Thanks a lot !!!!