Bourne shell script traverse

kdyzsa · 05-05-2008, 10:36 PM

I've only started learning UNIX 2 days ago since my supervisor wanted me to. She gave me some exercises which I successfully answered thanks to google and UNIX tutorials. But this one is taking me longer than usual so I come here to seek for help.

This is the exercise she gave me:
>> User will prompted to search for either by last name (-l) or first name(-f). The script named as searchcomp will get the exact matching parameter based from input.txt and display the breakdown of other details.
Format of execution will be like below:
searchcomp [-l] [-f] ‘pattern’
ex. searchcomp -f 'Tang'

Output will be:

First Name: <first name field>
Last Name: <last name field>
Exercise 1: <grade of exercise 1>
Exercise 2: <grade of exercise 2>
Case Study: <grade of case study>
Average: 15% of exer1 + 15% of exer2 + 70% of case study
Remark: ‘Fail’ for 74 below
‘Pass’ for 75 above
<<
Input.txt looks like this:
>>
LastName:FirstName:Exer1:Exer2:Case:
Tang:Sam:90:40:40
Dylan:Tom:80:48:50
Smith:John:34:67:89
...
<<
I don't have any problems with getting the average or whether its pass or fail. But I'm kinda stumped on how to get the firstname. My idea was to grep it then I really don't know how to seperate them since I can't put it in a seperate file. I'm nto allowed. So if the results of the grep were.
Ex.

Tang:Sam:90:40:40
Tang:Kim:94:78:40

I'm not sure how I can seperate the Tang as the lastname and Sam as a firstname? or the exercises? Is there any way to traverse the line/string? I'm also thinking if there were multiple results, I would have to seperate them as well. Am I on the right path or is my algorithm for solving this wrong? Any help would be appreciated with this, thanks

rocket357 · 05-05-2008, 10:46 PM

The "cut" command is what you're looking for...

Basic example given the following data in a file "input.txt":

Tang:Sam:90:40:40
Dylan:Tom:80:48:50
Smith:John:34:67:89

Find Sam Tang's first name only:

grep Tang input.txt | cut -d':' -f2

grep pulls just the line you want, then cut is passed two parameters: -d (delimiter, which in this case is set to ':') and -f (field, which in this case is "field" #2). If I used this command on /etc/passwd in this form:

snippet from /etc/passwd:

Code:

root:x:0:0:root:/root:/bin/bash
bin:x:1:1:bin:/bin:/bin/false
operator:x:11:0:operator:/root:/bin/bash

grep root /etc/passwd | cut -d':' -f1-2 --output-delimiter=" "

I get this:

root x
operator x

Edit: You can also accomplish this with sed or awk...I'll give a basic sed example (awk would be much easier!):

sed -e 's/$[A-Za-z]*$:$[A-Za-z]*$:$[0-9]*$:$[0-9]*$:$[0-9]*$/First Name: \1\nLast Name: \2\nExercise 1: \3\nExercise 2: \4\nCase Study: \5\n/g' input.txt

Disillusionist · 05-06-2008, 02:18 AM

Awk will produce much quicker results than the alternative Shell script, especially if the input file is large.

Here is a partial awk example:

Code:

#!/bin/bash
if [ "$1" == "-l" ]
then
awk -F":" 'function REPORT() {
   print "First Name:", $2
   print "Last Name :", $1
   print "Exercise 1:", $3
   print "Exercise 2:", $4
   print "Case Study:", $5
   l_avg=(($3 * .15) + ($4 * .15) + ($5 * .7))
   print "Average   :", l_avg
}
{
   if ( $1 == lName ) REPORT()
}' lName=$2 input.txt
fi

unSpawn · 05-06-2008, 06:56 AM

Also w/o awk/grep/cut, see BaSH pattern replacement: string="Tang:Sam:90:40:40"; string=("${string//:/ }"); echo "${string[0]}".

matthewg42 · 05-06-2008, 07:48 AM

If it's a shell scripting exercise for th sake of learning, using awk is probably not the right direction, although I think it's a good choice for doing such a task "for real". Perl nice - especially using the Getopt module provides a lot of functionality with a friendly interface (plus you can document your program with POD tags).