Shell Programming and Scripting

I have an inventory program that I would like to have the ability to go and change or alter the field data based on the item number as a key. I have the menu option set but at the end of the script process it just appends the changed data to the database rather than what I would like; which is to replace the current item number and fields that were changed in the menu.

Any advice would be appreciated.

The inventory change script is below.

#!/bin/bash
#===================================================
# Script Name: invadd
# By: jhaynes
# Date: 4-4-2006
# Purpose: Script loop to add inventory to the HC
# computing inventory list, while preventing
# duplicate item#.
# Command Line:invadd
#===================================================
trap "rm ~/tmp/* 2> /dev/null ; exit" 0 1 2 3
invfile=~/final/hc_inventory
looptest=y
while [ $looptest = "y" ]
do
clear
tput cup 25 4; ./appendinv;
tput cup 1 4; echo "HC Inventory Change"
tput cup 2 4; echo "================================"
tput cup 4 4; echo "Item Number: "
tput cup 5 4; echo "Description: "
tput cup 6 4; echo "Dept : "
tput cup 7 4; echo "Cost : "
tput cup 8 4; echo "Qty : "
tput cup 10 4; echo "Add Another? (y)es or (q)uit: "
tput cup 11 4; echo "Please enter an item number or q to quit "
tput cup 4 16; read itemnum
if [ "$itemnum" = 'q' ]
then clear ; exit
fi
#Check to make sure the Item number field is not blank
if [ "$itemnum" = "" ]
then
tput cup 11 2; echo "***Item number can not be blank*** "
tput cup 13 2; echo "******Hit ENTER to continue******"
tput cup 4 16 ; echo " "
tput cup 4 16 ; read itemnum
continue
fi
tput cup 5 16; read descr
while [ "$descr" = '-' ]
do
tput cup 4 16 ; echo " "
tput cup 4 16 ; read itemnum
if [ "$itemnum" = 'q' ]
then
clear ; exit
fi
tput cup 5 16 ; read descr
done
tput cup 6 16; read dept
while [ "$dept" = '-' ]
do
tput cup 5 16 ; echo " "
tput cup 5 16 ; read descr
if [ "$descr" = 'q' ]
then
clear ; exit
fi
tput cup 6 16 ; read dept
done
tput cup 7 16 ; read cost
while [ "$cost" = '-' ]
do
tput cup 6 16 ; echo " "
tput cup 6 16 ; read dept
if [ "$dept" = 'q' ]
then
clear ; exit
fi
tput cup 7 16 ; read cost
done
tput cup 8 16 ; read qty
while [ "$qty" = '-' ]
do
tput cup 7 16 ; echo " "
tput cup 7 16 ; read cost
if [ "$cost" = 'q' ]
then
clear ; exit
fi
tput cup 8 16 ; read qty
done
#Check to make sure fields are not blank before you write the data4
if [ "$itemnum:$descr:$dept:$cost:$qty" > " " ]
then
echo "$itemnum:$descr:$dept:$cost:$qty" >> $invfile
fi
tput cup 10 33; read looptest
if [ "$looptest" = 'q' ]
then
clear ; exit
fi
done

Editing a line in the middle of a file is a LOT harder than adding a line to the end of a file. Adding to the end just means opening in append mode, but selectively editing a line in the middle of it means reading in the file completely, modifying the contents, then writing back the edit and everything after it.

The reason it's so hard is, what if the length of the line changes? You can't just replace the line or stuff like this happens:

Code:

line1
line2
line3

Replace line2 with line99 and it becomes:

Code:

line1
line99line3

Because line99 is one character longer, 9 overwrites the newline character. To do this properly you have to insert more characters into the middle. Or, if the new line is shorter, delete characters from the middle. This makes it a real pain.

If you use fixed record sizes, you'll be able to just seek to and overwrite arbitrary records. See 'man dd'. bash may not be the best language for this sort of work, however.

Thanks Corona688. I understand what you are saying. I am useing fixed-length records. I will look to the man dd pages. To see if there is a way to update the veriables based on the item number being the key identifier to the line.

You are? Neat. I didn't catch that since I'm not familiar with all the commands you're using in your script.

dd doesn't give much consideration to what data it's copying or writing. You can't tell it what to overwrite, only where. Overwriting a specific record would be something like

Code:

echo "${DATA}" | dd bs=recordsize seek=recordnum of=recordfile

and reading a specific record would be like

Code:

DATA=`dd bs=recordsize seek=recordnum if=recordfile`

So you'll have to use other means to find out the record number.

Be aware that dd is not a bash builtin, running dd means creating a process.

I like to think of myself as neat also
Here is an example of the inventory list

HC Inventory List

Dept Item# Description Qty Cost
==== ===== ============================== ===== ======
0100 10001 Intel Celeron - 1.8GHz 7 $65.00
0100 10002 Intel Pentium 4 - 2.4GHz 10 $125.00
0200 10003 AMD Athlon XP 2800 12 $108.00
0100 10004 Intel Mobile Pentium 4 1.7GHz 6 $170.00
0100 10005 Intel Xeon 1.8GHz 8 $183.00
0100 10006 Intel Pentium 4 - 3.2GHz 4 $285.00
0300 10007 Kingston 256MB PC133 20 $50.39
0300 10008 Kingston 512MB RDRAM Kit 6 $240.00

So as an example, lets say I wanted to change the qty of item # 10007
other than dd, are there any other tools that might better achive what I am looking to do.

I forsee problems no matter what method you use. What you seem to be trying to build is a rudimentary relational database, to let you do queries like

Code:

UPDATE INVENTORY SET B="value1", C="value2", D="value3" WHERE A="12345";

but BASH is not a good language to build a relational database in. Bash The Database is a BASH extension that seems closer to what you want but is designed for connecting to remote databases, not local file-based ones. The BASH shell is just not designed for record-based file access, and neither are most scripting languages. Can you use other languages or do you absolutely need to use BASH for some reason?

Another thing I notice is that, unless you keep all the records in your database sorted by item number, there is no way to find a specific number except checking every single row. This will be very slow when your database grows larger. Keeping them ordered will let you find the record you want with a binary search.