Very Computer

I am writing a payroll script that accepts two command line parameters, two
times, starttime and endtime..
ie.  $ payroll 09:30 14:25

The script starts with an error checking if statement designed to make sure
the user inputs the times in proper format, HH:MM

This is the line I have right now, but I keep getting an error message

if [ "$1" = [0-2][0-9][:][0-5][0-9] -a "$2" = [0-2][0-9][:][0-5][0-9] ]

Using a 24 hour clock that's why my range goes from 00:00 to 23:59
The error message I keep getting says
[: too many integers

Does anyone have any idea what I'm doing wrong? Or does anyone have a better
solution to do this format check?

Darren

AFAIK, the [ command will not match with regexs.  You'll need to
either do something with grep, or use perl.

Another option would be to just pass the times to `date -d', if you
have gnu date.  Then the time can be in any common format, and it will
convert it to whatever format you ask for, so you can use a convenient
one (seconds since epoch seems useful)

Example:

starttime=`date -d "$1" +%s`
endtime=`date -d "$2" +%s`
minutes=`expr  \( $endtime - $starttime \) / 60 `
echo You worked for $minutes minutes.

--

Nate Eldredge

Quote:> Another option would be to just pass the times to `date -d', if you
> have gnu date.  Then the time can be in any common format, and it will
> convert it to whatever format you ask for, so you can use a convenient
> one (seconds since epoch seems useful)

> Example:

> starttime=`date -d "$1" +%s`
> endtime=`date -d "$2" +%s`
> minutes=`expr  \( $endtime - $starttime \) / 60 `
> echo You worked for $minutes minutes.

I can't use date because there is no guarantee they will enter a time at
all. If they enter some words as parameters instead of times I need my
script to tell them they didn't execute the program correctly.

Using perl is also not an option either unfortunately. Can you suggest how I
would use grep to perform this error check?

Darren

The '[' (or test) command doesn't "do" regular expressions. However,
case statements do (at least in modern shells). The following should do
what you want. As a general rule, case statements are always better for
string handling than 'test'.

for i in $1 $2; do
  case $i in
    [0-1][0-9]:[0-5][0-9]|[0-2][0-4]:[0-5][0-9]) echo "Valid time - $i"
;;
    *) echo "Invalid time - $i."
       exit 1 ;;
  esac
done

The above has the added advantage of not allowing times such as 29:00 to
slip through. :-)

Cheers,

Dave. :-)

You can echo the string to grep and redirect the output to /dev/null,
checking the exit value of grep, but a nicer way is to use the widely
available "expr" command's colon operator like so:

  if [ `expr "$1" : [0-2][0-9][:][0-5][0-9]` = 0 ]; then
      echo bad start time 1>&2
      exit 1
  fi
  if [ `expr "$2" : [0-2][0-9][:][0-5][0-9]` = 0 ]; then
      echo bad end time 1>&2
      exit 1
  fi

--
--Ed Cashin                     PGP public key:

    Note: If you want me to send you email, don't munge your address.

date will do that for you.

$ date -d "This is not a date"
date: invalid date `This is not a date'
$ echo $?
1

Then if the value of the variable is null, you know date failed and
you can barf appropriately.

Quote:> Using perl is also not an option either unfortunately. Can you suggest how I
> would use grep to perform this error check?

echo $user_input | grep "time_regex" || echo Sorry, bad date ; exit 1

where time_regex is your carefully crafted regex that matches a time.

--

Nate Eldredge

Right.  [0-2][0-9][:][0-5][0-9] and [0-2][0-9][:][0-5][0-9] are probably
numerous test constructs as far as the shell (Bash, and POSIX I think) are
concerned. You didn't even say what kind of script, so I assume a sh. sh
does filename globbing itself, but not regexes of arbitrary strings. Like
Perl, Awk and the Plan9 shell `rc' have "compare regex against string",
but "compare regex to regex" sounds like something that might be provably
impossible or something anyway.

"[: too many integers"   is probably failing at the first [ in "[0-2]".

The shell is very powerful, and thus confusing. You have to write it one
character at a time as pertains to the test command, in my experience.
I'd never imagine a big test construct like yours working on the first go.
You need to proceed in smaller steps. K.I.S.S.

Rick Hohensee

"So far, Linux doesn't have a lot of traction on the client
[Microsoft-ese for desktop computers], except in some university
environments."
Steve Ballmer of Microsoft, as quoted and remarked by John Schwartz
in the Washington Post, June 11 2000